Upload
dongoc
View
215
Download
0
Embed Size (px)
Citation preview
General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
• Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from orbit.dtu.dk on: Mar 15, 2018
Monitoring of priority pollutants in dynamic stormwater discharges from urban areas
Birch, Heidi; Mikkelsen, Peter Steen; Lützhøft, Hans-Christian Holten
Publication date:2012
Document VersionPublisher's PDF, also known as Version of record
Link back to DTU Orbit
Citation (APA):Birch, H., Mikkelsen, P. S., & Lützhøft, H-C. H. (2012). Monitoring of priority pollutants in dynamic stormwaterdischarges from urban areas. Kgs. Lyngby: DTU Environment.
PhD ThesisAugust 2012
Heidi Birch
Monitoring of priority pollutants in dynamic
stormwater discharges from urban areas
Monitoring of priority pollutants in dynamic
stormwater discharges from urban areas
Heidi Birch
PhD Thesis
August 2012
DTU Environment
Department of Environmental Engineering
Technical University of Denmark
DTU Environment
August 2012
Department of Environmental Engineering
Technical University of Denmark
Miljoevej, building 113
DK-2800 Kgs. Lyngby
Denmark
+45 4525 1600
+45 4525 1610
+45 4593 2850
http://www.env.dtu.dk
Vester Kopi
Virum,
Torben Dolin
978-87-92654-72-4
Address:
Phone reception:
Phone library:
Fax:
Homepage:
E-mail:
Printed by:
Cover:
ISBN:
Heidi Birch
Monitoring of priority pollutants in dynamic stormwater discharges from
urban areas
PhD Thesis,
The thesis will be available as a pdf-file for downloading from the homepage of
the department: www.env.dtu.dk
August 2012
i
Preface This thesis presents the outcome of a PhD project carried out at the Department
of Environmental Engineering (DTU Environment), Technical University of
Denmark (DTU) in the period from January 2008 to June 2012. The project was
supervised by Associate Professor Peter Steen Mikkelsen and Associate
Professor Hans-Christian Holten Lützhøft.
The thesis is based on five scientific papers. In the thesis, these papers are
referred to with the roman numbers (e.g. Paper I).
I. Birch, H., Mikkelsen, P.S., Jensen, J.K., Lützhøft, H.-C.H. (2011).
Micropollutants in stormwater run-off and combined sewer overflow in
the Copenhagen area, Denmark, Water Science and Technology, 64 (2),
485-493. DOI: 10.2166/wst.2011.687.
II. Birch, H., Gouliarmou, V., Lützhøft, H.-C.H., Mikkelsen, P.S. and
Mayer, P. (2010). Passive Dosing to determine the speciation of
hydrophobic organic chemicals in aqueous samples, Analytical Chemistry,
82 (3), 1142-1146. DOI: 10.1021/ac902378w.
III. Birch, H., Mayer, P., Lützhøft, H-C.H. and Mikkelsen, P.S. (in revision)
Partitioning of fluoranthene between free and bound forms in stormwater
runoff and urban waste waters using passive dosing.
IV. Birch, H., Sharma, A.K., Vezzaro, L., Lützhøft, H.-C.H. and Mikkelsen,
P.S. (manuscript). Velocity dependant sampling of micropollutants in
stormwater using a flow-through passive sampler.
V. Birch, H., Vezzaro, L. and Mikkelsen, P.S. (accepted) Model based
monitoring of stormwater runoff quality. In proceedings of the 9th
international conference on Urban Drainage Modelling, Belgrade, 2012.
The papers are included in the printed version of the thesis but not in the www
version. Copies of the papers can be obtained from the Library at DTU
Environment.
ii
Department of Environmental Engineering
Technical University of Denmark
Miljoevej, Building 113
DK-2800 Kgs. Lyngby, Denmark
The following articles and reports were prepared during the PhD project but they
are not included as part of the thesis:
Birch, H., Mikkelsen, P.S., Jensen, J.K. and Lützhøft, H.-C.H. (2010).
Xenobiotics in Stormwater Run-off and Combined Sewer Overflow. Proc. 14th
Int. Conf. on Diffuse Pollution and Eutrophication – DIPCON 2010, September
12-17, Baupré, Quebec, Canada, Abstract (S901-IWA-4679R1, p. 124) and Full
paper (S901, p. 247-252).
Birch, H., Sharma, A.K., Vezzaro, L., Mikkelsen, P.S. and Lützhøft, H.-C.H.
(2011). Passive sampling and modeling of PAHs and heavy metals in stormwater
runoff, In Proceedings of the12th International Conference on Urban Drainage,
Porto Alegre/Brazil, 11-16 September 2011. Full paper PAP005248. 8 p.
Birch, H., Lützhøft, H.C.H. and Mikkelsen, P.S. (2010). Tungmetaller og
miljøfremmede organiske stoffer i regnbetingede udledninger: nye
stikprøvemålinger (Heavy metals and xenobiotics in stormwater discharges: new
spot sample measurements), danskVAND (Danish Water), 78 (5), 37-40.
Fryd, O., Backhaus, A., Jeppesen, J., Ingvertsen, S., Birch, H., Bergman, M.,
Panduro, T., Fratini, C., Jensen, M.B. (2009). Koblede afkoblinger. Vilkår for
landskabsbaserede afkoblinger af regnvand i det københavnske kloakopland til
Harrestrup Å.: Arbejdsrapport (Connected disconnections. Conditions for
landscape based stormwater handling in the Copenhagen sewer catchment to
Harrestrup Å: Working report). By&Landskab, Københavns Universitet. 71 s.
http://2bg.dk/Internal_Workshop/2009-12-03-
KE/2BG_HarrestrupAa_Booklet_web.pdf
Fryd, O., Backhaus, A., Birch, H., Fratini, C., Ingvertsen, S.T., Jeppesen, J.,
Panduro, T.E., Roldin, M.K., Dam, T.E., Wenningsted-Torgard, R.M. and
Jensen, M.B. (2012). Potentials and limitations for Water Sensitive Urban Design
in Copenhagen: a multidisciplinary case study. Paper presented at the 7th
International Conference on Water Sensitive Urban Design, Melbourne,
Australia, 21-23 february 2012
iii
Jensen, M.B., Fryd, O., Backhaus, A., Ingvertsen, S.T., Pauleit, S., Dam, T.E.,
Bergman, M., Birch, H. and Mikkelsen, P.S. (2009). Holland: Landskabsbaseret
afvanding (The Netherlands: Landscape based drainage). Teknik & Miljø: Stads
og Havneingeniøren. 38-40.
Lützhøft, H.-C.H., Birch, H., Eriksson, E. and Mikkelsen, P.S. (2011).
Comparing chemical analysis with literature studies to identify micropollutants in
a catchment of Copenhagen (DK). In Proceedings of the 12th International
Conference on Urban Drainage, Porto Alegre, Rio Grande do Sul, Brazil, 11th-
16th September 2011. Full paper: PAP005319. 8 p.
Pettersson, M., De Keyser, W., Birch, H., Holten Lützhøft, H.-C., Gevaert, V.,
Cerk, M., Benedetti, L., and Vanrolleghem, P.A., (2010). Strategies for
Monitoring of Priority Pollutant Emission Barriers. ScorePP project deliverable
D7.5, 66 p. (available at www.scorepp.eu).
Sharma, A.K., Vezzaro, L., Birch, H. and Mikkelsen, P.S. (2011). Effect of
climate change on stormwater characteristics and treatment efficiencies of
stormwater retention ponds. In Proceedings of the 12th International Conference
on Urban Drainage, Porto Alegre, Rio Grande do Sul, Brazil, 11th-16th
September 2011.Full paper: PAP005314. 8 p.
v
Acknowledgements I would like to thank my supervisors Peter Steen Mikkelsen and Hans-Christian
Holten Lützhøft for providing valuable support, encouragement and helpful
advice throughout the PhD study. I am also grateful to Niels Bent Johansen and
Jeanet Stagsted Nielsen from Copenhagen Energy for inspiring discussions and
comments.
The project was funded by DTU, Copenhagen Energy (KE) and the Ministry of
Science, Technology and Innovation through the Urban Water Technology
Graduate School. The results were obtained with additional financial support
from
- The Danish Council for Strategic Research through the project 2BG:
Black Blue Green, Integrated Infrastructure Planning as Key to
Sustainable Urban Water Systems (www.2BG.dk), and the REMTEC
project.
- The EU Interreg IV B North Sea Regional Programme through the project
“Diffuse Pollution” (DIPOL).
- The Interreg IIIB Baltic Sea Programme through the project COHIBA -
Control of hazardous substances in the Baltic Sea Region.
- The European Commission through the project: Source Control Options
for Reducing Emissions of Priority Pollutants (ScorePP), funded under the
Sixth Framework Programme (www.scorepp.eu) and the project OSIRIS,
COGE-037017.
- Albertslund Municipality.
- COST action 636 “Xenobiotics in the urban water cycle” (support for
conference attendance).
- Otto Mønsteds Fonden (support for conference attendance).
I would like to thank my PhD colleagues from the 2BG project, Ole Fryd, Antje
Backhaus, Jan Jeppesen, Maria Roldin, Simon Toft Ingvertsen, Toke Panduro
and Chiara Fratini for fruitful collaboration and Marina Bergen Jensen for
coordinating the effort of all of us in this project. I would also like to thank the
partners from the ScorePP project for interesting discussions and collaboration.
vi
I would like to thank the partners in the Dipol project, Hans-Henrik Høg from
Albertslund municipality for allowing me to install equipment in their
stormwater system, Thomas Aabling from Orbicon A/S for advice and practical
help with the installations, Anitha Sharma from DTU Environment for
collaboration during the project and Luca Vezzaro from DTU Environment for
letting me use his model and for help and assistance with the modeling work.
I am very grateful to Christel Sebastian for providing samples from the retention
pond in Chassieu and for measuring TSS and loss on ignition on those samples
for me.
I am also very grateful to my co-authors Philipp Mayer and Varvara Gouliarmou
from Department of Environmental Science, Aarhus University and Jørgen
Krogsgaard Jensen from DHI. It was a pleasure working together with you.
I would like to thank the laboratory technicians at DTU Environment, especially
Mikael Emil Olsson and Sinh Nguyen for their help and assistance in the
laboratory.
I thank my colleagues and office-mates who inspired me and cheered me up. My
appreciations go to Anne Harsting for her support on all the administrative
matters.
Tak til Daniel, Casper og Lukas for hygge og glæde i fritiden.
vii
Abstract The European Water Framework Directive (WFD) from 2000 has put focus on
the chemical status of surface waters by the specified Environmental Quality
Standard (EQSs) and the requirements for monitoring of surface water quality
throughout Europe. When considering the water quality of urban stormwater
runoff it is evident that surface waters receiving large amount of urban
stormwater runoff will be at risk of failing to meet the EQSs. Therefore
stormwater treatment is crucial. However, as stormwater quality varies orders of
magnitude between sites, stormwater monitoring is important in order to design
the right treatment level to protect surface waters. Stormwater runoff is very
dynamic both quality and quantity wise. In order to optimize the sampling of
such phenomena, advanced sampling equipment is required. Such equipment is
expensive, and furthermore, it is time consuming to conduct the sampling
campaigns. Therefore this PhD project aimed at improving monitoring programs
for priority pollutants in stormwater runoff.
By comparing results from a literature study and a screening campaign to the
EQSs, it was found that heavy metals (especially Cu and Zn), polyaromatic
hydrocarbons (PAHs), Di(2-ethylhexyl)-phthalate (DEHP) and pesticides were
the main pollutants of general concern in stormwater runoff and of concern at the
studied catchment (glyphosate was found to be the most relevant pesticide in a
Copenhagen setting). These priority pollutants are therefore relevant to monitor
in stormwater discharges.
Sorption of pollutants to particulate matter and dissolved organic carbon is
important for both the toxicity of the pollutants and for removal in stormwater
treatment systems. Furthermore sorption is important for sampling using the most
common types of passive samplers, which are based on uptake of analytes by
diffusion, since they only sample the freely dissolved and labile fraction of
analytes. Passive dosing was therefore developed during this PhD project as an
easy, fast and precise method for partition measurements of hydrophobic organic
compounds (HOC) in aqueous samples such as stormwater runoff. The principle
of the method is that the freely dissolved concentration of the HOC is controlled
by partitioning from a pre-loaded polymer and the total concentration in the
sample at equilibrium is measured. Partition measurements in stormwater runoff
samples revealed a partition ratio log KTSS for fluoranthene of 4.59, and free
viii
fractions in stormwater runoff of 0.04-0.5. The partition ratio can be used in
modeling of stormwater treatment systems. The passive dosing method can be
used for surface water monitoring to relate freely dissolved concentrations to
total concentrations.
For stormwater monitoring, diffusion based passive samplers are not appropriate
to use. The reason is that the sampler measures time-weighted concentrations
over periods of weeks to months with no regard to whether it rains or not.
Therefore a flow-through passive sampler, SorbiCell, was tested. It consists of a
cartridge containing a sorbent and was installed directly in the stormwater
drainage ditch letting the momentum from the water velocity force water through
the sampler. This novel installation method ensures sampling mainly during
runoff events and dependant on the velocity of the runoff. Even though a filter
prevented large particles from entering the sampler, it revealed concentrations
comparable to volume proportional total concentrations measured in stormwater
runoff and modeled using a dynamic stormwater quality model. There are still
many questions and assumptions when using this installation method. However it
has potential for monitoring the load of priority pollutants to surface waters from
the large amounts of stormwater discharge points often contributing to the
deterioration of the water quality.
When evaluating the pollutant level at specific sites based on measurements, an
interpretation of the system is always involved. This interpretation can be
formulated in stormwater quality models. Event mean concentrations (EMCs) are
often found to follow a lognormal distribution. However more complicated
models including dynamics of accumulation in the catchment and influence of
rain characteristics on the runoff concentrations can also be used. The advantage
of using models for monitoring purposes is that information about the system
beyond the time interval of sampling can be obtained based on knowledge of
processes and observed patterns. It was found here that model prediction bounds
for annual average concentrations obtained by a dynamic stormwater quality
model were narrower than uncertainty on the mean when assuming lognormal
distribution of EMCs. Furthermore, the use of passive sampler measurements in
combination with volume proportional measurements for calibration reduced the
model prediction bounds on annual average concentrations more than simply
increasing the number of volume proportional samples. This work demonstrated
how models and passive samplers can be used for monitoring purposes.
ix
Dansk Sammenfatning Vandrammedirektivet fra år 2000 har sat fokus på den kemiske vandkvalitet af
overfladevand ved at definere vandkvalitetskriterier og krav om overvågning af
kvaliteten af overfladevand i Europa. Det er imidlertid klart når man ser på
kvaliteten af regnvandsafstrømning, at overfladevand der tilføres store mængder
regnvandsafstrømning fra byområder vil risikere ikke at kunne overholde disse
kvalitetskrav. Derfor er rensning af regnvandsafstrømning vigtig. Men eftersom
kvaliteten af regnvandsafstrømning varierer størrelsesordner mellem oplande, er
målekampagner vigtige for at kunne finde det rigtige rensningsniveau til at
beskytte overfladevandet. Regnvandsafstrømning er meget dynamisk både mht.
kvalitet og kvantitet. Derfor er avanceret prøvetagningsudstyr nødvendigt. Dette
er dyrt, og prøvetagning er derudover knyttet til et højt tidsforbrug. Formålet med
denne PhD var derfor at forbedre prøvetagningsprogrammer for miljøfremmede
stoffer i regnvandsafstrømning.
Ved at sammenligne resultater fra litteraturen og en screeningsmålekampagne
med kvalitetskriterierne, blev det konkluderet at tungmetaller (specielt Cu og
Zn), polyaromatiske hydrocarboner, Di(2-ethylhexyl)-phthalate (DEHP) og
pesticider er de primære problematiske miljøfremmede stoffer generelt i
regnvandsafstrømning såvel som i de oplande der indgik i analysen (glyphosat
var det mest relevante pesticid i Københavnsområdet). Disse stoffer er derfor
relevante at overvåge i regnvandsafstrømning.
Sorption af miljøfremmede stoffer til partikler og opløst organisk materiale er
vigtigt for både toksiciteten af stoffet og fjernelsen i bassiner og andre metoder til
rensning af regnvand. Derudover er sorption vigtig når de mest udbredte typer af
passive samplere benyttes. De optager nemlig stoffer ved diffusion, og optager
derfor kun den totalt opløste del. Passive dosing blev udviklet som en enkel,
hurtig og præcis måde til at måle fordeling af hydrofobe organiske stoffer i
vandige prøver så som regnvandsafstrømning. Princippet i metoden er at den frit
opløste koncentration af stoffet bliver styret af en ligevægt med en polymer der
indeholder stoffet. Den totale koncentration i prøven ved ligevægt bliver
efterfølgende målt. Fordelingsmålinger i regnvandsafstrømning viste en
fordelingsligevægt, log KTSS, for fluoranthene på 4.59, og frit opløste
koncentrationer i regnvandsafstrømning på 4 - 50%. Passive dosing metoden kan
x
bruges til overvågningsprogrammer for overfladevand når koncentration af frit
opløste stoffer skal relateres til totale koncentrationer.
Passive samplere som er baseret på diffusion er imidlertid ikke velegnede til at
bruge til overvågning af regnvandsafstrømning. Det skyldes det faktum at denne
type passive samplere måler tidsvægtet over tidsperioder på uger til måneder
uden hensyn til om det rent faktisk regner eller ej. Derfor blev en passiv sampler
der opsamler stof som funktion af vandgennemstrømningen, SorbiCell, testet.
Den består af et hylster med en sorbent i til at opfange stoffer når vand passerer
gennem hylsteret. Den blev installeret direkte i regnvandsgrøften på en sådan
måde at vandhastigheden af det strømmende vand pressede vand gennem
sampleren. Denne metode sikrer prøvetagning primært ved afstrømning og
afhængig af vandets hastighed. Selv om et filter forhindrede store partikler i at
komme ind i sampleren, målte den passive sampler koncentrationer svarende til
totale koncentrationer i volumen proportionelle prøver af
regnvandsafstrømningen. Der er stadig mange spørgsmål og antagelser når den
passive sampler bruges med denne installationsmetode. Den har imidlertid et
stort potentiale ved overvågning af belastningen af overfladevand med
miljøfremmede stoffer fra regnvand fra en stor mængde udledningspunkter og
oplande.
Når koncentrationsniveauet et specifikt sted skal vurderes baseret på målinger, er
det altid baseret på en bestemt forståelse af systemet. Denne forståelse kan
formuleres i modeller af kvaliteten af regnafstrømning. Hændelsesmiddel-
koncentrationer følger ofte en lognormal fordeling. Mere komplicerede modeller,
som tager højde for akkumulation af stofferne på overflader og indflydelsen af
regnintensitet og varighed på koncentrationen i afstrømningen, kan også
benyttes. Fordelen ved at bruge modeller til overvågning er at de giver
information om systemet ud over den aktuelle måleperiode baseret på viden om
processer og observerede mønstre. I dette arbejde var usikkerhedsintervallet på
årsgennemsnittet mindre når en dynamisk afstrømningsmodel blev brugt end
usikkerheden på gennemsnittet af en lognormal fordeling af
hændelsesmiddelkoncentrationer. Brug af målinger med passive samplere
kombineret med volumen proportionelle prøver til kalibrering af modellen
reducerede usikkkerhedsintervallet på årsgennemsnittet mere end ved at bruge en
større mængde volumen proportionelle prøver til kalibrering. Dette arbejde viste
hvordan modeller og passive samplere kan bruges i overvågningen
xi
Table of contents
1 Introduction .................................................................................................... 1
1.1 The challenge of monitoring priority pollutants in stormwater runoff ..... 1
1.2 Aims and research questions ..................................................................... 3
1.3 Methods and overview of thesis ............................................................... 4
1.4 Field work: locations and sampling .......................................................... 6
2 Monitoring ...................................................................................................... 9
2.1 Water Framework Directive ..................................................................... 9
2.2 Monitoring priority pollutants in stormwater runoff .............................. 11
3 Priority pollutants in stormwater ............................................................... 13
3.1 Selection of pollutants for monitoring .................................................... 13
3.2 Sources and occurrence of pollutants in stormwater .............................. 15
4 Partitioning of priority pollutants in stormwater ..................................... 21
4.1 General principles ................................................................................... 21
4.2 Methods for speciation and partitioning measurements ......................... 22
4.3 Partitioning of PAHs ............................................................................... 24
5 Sampling priority pollutants in stormwater .............................................. 29
5.1 Traditional methods ................................................................................ 29
5.2 Passive sampling ..................................................................................... 31
5.2.1 Diffusion based passive samplers .................................................... 31
5.2.2 Flow-through passive samplers ........................................................ 35
6 Monitoring of priority pollutants in stormwater runoff using passive
samplers and models .................................................................................... 39
6.1 Modeling stormwater quality .................................................................. 41
6.2 Use of a dynamic stormwater runoff quality model for evaluating
monitoring strategies ............................................................................... 43
7 Conclusions ................................................................................................... 47
8 Future research ............................................................................................ 49
9 References ..................................................................................................... 51
10 Papers ............................................................................................................ 61
1
1 Introduction 1.1 The challenge of monitoring priority pollutants in
stormwater runoff Increased urbanization has many effects on the environment. One of these is the
impact on the natural water cycle. This includes impacts on the quantity and
quality of the water reaching the lakes and streams. In the beginning of the last
century the largest impacts on the quality of urban surface waters was caused by
untreated wastewater being discharged directly. As wastewater treatment plants
are now handling wastewater in urban areas in Europe focus has changed to
discharges of stormwater (also often called stormwater runoff) and combined
sewer overflows which, during rain, contribute substantially to the pollution of
surface waters (Clark et al., 2008; Eriksson et al., 2007b; Göbel et al., 2007;
Kjølholt et al., 2007).
In Europe, the water quality of surface waters has been brought into focus by the
Water Framework Directive (WFD) from 2000 (European Commission, 2000).
The Water Framework Directive aims at improving the ecological status of
inland and coastal waters by 2015. This includes reducing the pollution of
surface waters with 45 identified priority substances and phasing out emissions,
discharges and losses of 17 of these, which are identified as priority hazardous
substances. For these substances environmental quality standards (EQS) have
been defined as annual average (AA-EQS) and maximum allowable
concentrations (MAC-EQS) (European Commission, 2008). These quality
standards are regularly updated and a new version including 15 additional
priority substances is presently proposed (European Commission, 2011). For
heavy metals the EQSs are defined for dissolved concentrations, reflecting the
fact that this is the fraction of the pollutants responsible for the toxic effect.
However for organic pollutants, the EQSs are defined for total concentrations,
reflecting the lack of knowledge about partitioning of organic pollutants in
natural environments.
The EQSs are in Denmark implemented in the executive order 1022 (Danish
Ministry of the Environment, 2010). Compared to the WFD it includes more
substances and more stringent criteria for some of the substances. It is used for
discharge permits with the exception of discharges of ‘normally loaded’
stormwater. However, even though stormwater discharge permits are not
2
targeted, the WFD EQSs are indirectly applicable for stormwater discharges
since they apply to the recipients.
The WFD also includes requirements for monitoring of the ecological and
chemical status of the water bodies in all member states (European Commission,
2000). The monitoring includes the chemical status regarding priority substances
and other relevant pollutants (in this thesis the term priority pollutants, PPs, will
be used to denote all the relevant pollutants, organic compounds as well as heavy
metals, whether included in the WFD or not). Monitoring of stormwater is not
specifically required in the WFD. However, for many surface waters stormwater
is a major contributor of PPs. Therefore these discharges will have to be treated
in some way. In order to design the right level of treatment, monitoring of
stormwater is necessary (Ingvertsen et al., 2011).
The traditional methods for taking water samples include methods from manual
‘grab sampling’ to time, flow or volume proportional sampling using automatic
samplers. The samples are then taken to the laboratory for analysis, which can be
chemical analysis and/or toxicity tests. A range of alternative sampling methods
have been developed in the later years focusing on in situ extraction (bringing
only the extract/sorbent to the lab for analysis) or in situ measurements (see
Table 1). The use of these alternative sampling methods for monitoring purposes
are discussed and exemplified in Allan et al. (2006a; 2006b). Some methods are
appropriate for point measurements in time (instant measurements) while other
methods are suitable for time/volume/flow proportional sampling (Table 1).
Table 1: Classification of sampling methods according to whether the method target pollutant
concentrations or toxicity in samples, in situ analysis or laboratory analysis and instant or
average concentrations/effects.
Pollutant concentrations Toxicity measurements
Instant Average Instant Average
Extraction and analysis in lab
Grab sampling Time, volume and flow proportional sampling
Bioassays (performed on grab or time/flow proportional samples)
In situ extraction or measurement
Passive samplers (equilibrium)
Immunoassays
Sensors
Passive samplers (kinetic)
Biological early warning systems
Biomarkers
3
The advantage of using measurements of toxicity is that the combined effect of
all the pollutants is measured directly. However toxicity measurements alone do
not show which pollutants are responsible for the effects seen. The advantages of
pollutant concentration measurements are that legislation is based on chemical
concentrations and that the pollutants can be linked to sources. Furthermore
sources can be regulated and remediating actions can also be evaluated on the
basis of change in pollutant concentrations. The disadvantage is that important
pollutants can be ‘missed’ because they were not looked for and that combined
effects of multiple PPs are not taken into consideration.
The challenge when monitoring stormwater runoff is the very dynamic nature of
the system. Both the quantity and the quality of the stormwater vary orders of
magnitudes between different sites (Göbel et al., 2007; Maestre and Pitt, 2005).
Furthermore the variations over time at each location can also be huge (Lee et al.,
2007). This challenge has traditionally been approached by using flow- or
volume proportional sampling equipment. By sampling a number of different
rain events covering seasonal variations, a range of event mean concentrations
(EMCs) can be found. These EMCs can then be compiled into a single site mean
concentration (SMC) which can be used to evaluate the load of pollutants from
the discharge point (Mourad et al., 2005). This approach is both time consuming
and costly, especially if a broad range of PPs, for which chemical analysis can be
very expensive, have to be monitored. The many novel passive sampling
techniques have only to a small extent been used in stormwater drainage systems
(e.g. Paper IV, Blom et al. (2002) and Komarova et al. (2006)). Effort have also
been put in developing stormwater quality models for PPs in order to expand the
knowledge of the system beyond the actual sampled events (Vezzaro et al., 2012;
Vezzaro and Mikkelsen, 2012). However the role of passive samplers and
stormwater models for monitoring purposes has not yet been investigated.
1.2 Aims and research questions The aim of this thesis is to investigate the possibility to improve monitoring
programs for PPs in stormwater runoff. The main hypothesis of the thesis is that
a combination of active and passive sampling with stormwater quality modeling
can be used for monitoring concentrations and loads of PPs in stormwater in a
smarter way than the traditional monitoring programs. This will lead to lower
cost and/or deeper knowledge gained by the monitoring programs by including
more substances or sites in the program. This is greatly needed since the amount
4
of pollutants targeted in legislation is increasing whereas monitoring budgets are
strict or being cut in many countries.
The following research questions are studied in this thesis:
1. Which pollutants are relevant to monitor in stormwater?
2. How can the sorption capacity of stormwater be characterized in order to
relate total concentrations (targeted in regulation of organic substances) to
dissolved concentrations (responsible for toxic effects) and better predict
the fate of priority pollutants in stormwater treatment systems?
3. Can passive samplers be used for monitoring of stormwater discharges?
Which advantages, disadvantages and uncertainties do they have?
4. How can models in combination with new technological sampling
techniques improve monitoring of dynamic discharges from urban areas?
1.3 Methods and overview of thesis The methods used to approach these research questions included literature
studies, field work, development of an analytical method and stormwater quality
modeling. Figure 1 shows a conceptual drawing of a stormwater drainage system
where roof and road runoff is collected in storm sewers and treated in a retention
pond before discharge to a stream. The drawing shows how the stormwater
system was approached at different levels in this PhD project and thesis: overall
monitoring strategies involving the whole drainage system and surface waters
(section 2 and 6), evaluating pollution and modeling on a catchment level
(section 3 and 6), looking into specific sampling techniques for stormwater
(section 5), and focusing on partitioning of organic pollutants in stormwater
(section 4).
The thesis starts with a general presentation of different types of monitoring
defined in the WFD (Section 2).
In section 3, the occurrence and relevance of pollutants in stormwater is
investigated. This includes among others the results from a screening campaign
conducted in Copenhagen (Paper I).
5
Figure 1: Conceptual sketch of a stormwater drainage system.
An important characteristic of the fate of PPs is the sorption to particulate and
dissolved matter. Partitioning is discussed in Section 4. This includes a new
analytical technique for partitioning measurements of hydrophobic organic
compounds (HOCs), passive dosing, which was developed during the PhD
(Paper II). The importance of partitioning in stormwater runoff is also discussed
based on application of the method to stormwater runoff samples (Paper III).
Sampling of stormwater runoff is discussed in section 5. Specific focus is put on
passive sampling methods, including tests of a flow-through passive sampler
which was installed for velocity dependant sampling (Paper IV).
This is wrapped up by section 6, where the role of the passive samplers, passive
dosing and modeling for monitoring purposes is discussed. This section also
includes a discussion of stormwater modeling in general.
Diss. organic matter.
.... ..
. .... ..
Particles
Monitoring: Chapter 2 and 6PPs and modelling in stormwater:Paper I and V, Chapter 3 and 6
Partitioning: Paper II and III
Chapter 4
.........
Sampling: Paper IVChapter 5
6
1.4 Field work: locations and sampling Two main sampling campaigns were completed during the PhD project, an initial
screening campaign and an extensive sampling campaign in Albertlund.
Furthermore samples from a retention pond in Chassieu in France were used.
The screening campaign consisted of one grab/precipitation dependant/volume
proportional sample in six storm sewers (four different sites) and one CSO
(Table 2).
After the initial screening campaign and method development, field work was
focused in the catchment in Albertslund (Fabriksparken) where stormwater
runoff from an industrial/residential area is treated in a retention pond (basin K)
before discharge to a stream (Harrestrup Å). The events sampled during this
sampling campaign are shown in Figure 2.
Table 2. Description of the sites, samples and rain events studied in the initial screening
campaign (from Paper I).
Town Tårnby Tårnby Albertslund Glostrup Gentofte
Sites Byparken Digevej Fabriks-
parken
Ejby mose Scher-
figsvej
Sewer type Storm sewer Storm sewer Storm sewer Storm sewer CSO1
Impervious
area
1.3 ha 9.4 ha 56 ha 13 ha 60 ha 5 ha 1100 ha
Catchment
type2
Roads Res. and
metro
Ind. & res. Res. Ind. &
road
Res. &
road
Res. & roads
& drains
Treatment Grit
chamber
Grit
chamber
Oil sep.4
Oil
sep.4
Oil
sep.4
Oil
sep.4
-
Sample
type
Grab Grab Grab Precipitation dependant5 Volume
proportional
Date 10/15 2008 11/18 2008 11/18 2008 09/02 - 09/03 2009 09/30 2008
Rain depth6 2.3 mm 1 mm 5.7 mm 11 mm 6.4 mm
Duration6 3 hours
30 min 3 hours 16 h 23 h
Antecedent
dwp7
9 days 36 hours 36 hours 36 hours 7 days
1Combined Sewer Overflow,
2Res. = residential, Ind = Industrial,
3natural wetland,
4separator,
5Samples are taken depending on precipitation measured in a rain gauge at the site,
6For grab
samples the depth and duration refers to rain depth and how long time it had rained when the
grab sample was taken, not the total rain event depth and duration, 7dwp=dry weather period.
7
Figure 2: Flow during the extensive sampling campaign in Albertslund. The three events
sampled using autosamplers during spring 2010 and fall 2010 and the four events during spring
2011 are indicated. Two passive sampler installation periods are shown as well.
In total 10 events were sampled using autosamplers to sample volume
proportionally in the inlet to the pond and time proportionally in the outlet from
the pond. During two periods passive samplers were also installed here (see
Figure 2).
0
100
200
300
400
500
600
10/09 20/09 30/09 10/10 20/10 30/10 09/11
Flo
w, l
/s
Fall 2010
Event 2
Event 3Event 1
Passive sampler installation period
Passive sampler installation period
0
100
200
300
400
500
600
10/05 20/05 30/05
Flo
w, l
/sSpring 2010
Event 2
Event 3Event 1
0
100
200
300
400
500
600
29/03 08/04 18/04 28/04 08/05 18/05
Flo
w, l
/s
Spring 2011
Event 1
Event 4
Event 2
Event 3
8
The samples collected in Albertslund were used for multiple purposes.
Partitioning of fluoranthene in stormwater runoff were measured in samples from
this catchment (events 1, 3 and 4 from spring 2011), using the passive dosing
method (Paper III). Initial tests of a flow-through passive sampler were
performed during fall 2010, and during spring 2011 volume proportional
sampling were used in parallel with passive sampling in order to compare the two
methods (Paper IV). These samples were furthermore used to evaluate different
sampling strategies by applying a stormwater runoff quality model (Paper V).
Spring and fall events from 2010 were used to calibrate the model, and the four
events and passive samplers from spring 2011 were used to evaluate hypothetical
sampling strategies which could be adopted for further analysis of the catchment.
The retention pond in Chassieu in France receives stormwater from an industrial
catchment with an impervious area of 139 ha (Bertrand-Krajewski et al., 2008).
Partitioning of fluoranthene was measured in three composite stormwater
samples from the inlet and the outlet of the retention pond. The results are
compared to the results from Albertlund in this thesis (Section 4.3).
9
2 Monitoring 2.1 Water Framework Directive Monitoring of PPs in the environment is closely connected with requirements in
the legislation. Therefore, the substances included in the WFD, their AA-EQS
and MAC-EQS values and the requirements for monitoring are drivers for
establishing monitoring campaigns in Europe. In the WFD three types of
monitoring are defined: surveillance, operational and investigative monitoring.
The aims of these three types of monitoring are listed in Figure 3.
Aims of surveillance monitoring: - supplementing and validating the impact assessment, - efficient and effective design of future monitoring programmes, - assessment of long-term changes in natural conditions, - assessment of long-term changes resulting from widespread anthropogenic activity.
Aims of operational monitoring: - establish the status of those bodies identified as being at risk of failing to meet their environmental objectives - assess any changes in the status of such bodies resulting from the programmes of measures.
Investigative monitoring should be carried out: - where the reason for any exceedance is unknown, - where surveillance monitoring indicates that the objectives set out for a body of water are not likely to be achieved and operational monitoring has not already been established, in order to ascertain the causes of a water body or water bodies failing to achieve the environmental objectives, - to ascertain the magnitude and impacts of accidental pollution
Figure 3: The aims of the three types of monitoring programs as defined in the WFD (European
Commission, 2000).
Requirements for sampling stated in the WFD are that sampling should conform
to the ISO-guidelines and that for priority substances a frequency of one sample
per month is used as standard. For analysis, “the minimum performance criteria
for all methods of analysis applied are based on an uncertainty of measurement
of 50% or below estimated at the level of relevant EQS and a limit of
quantification equal or below a value of 30% of the relevant EQS” (European
Commission, 2009). The WFD opens up for variations in sampling methods,
frequencies and local considerations e.g. by saying that the frequency should be
chosen “so as to provide sufficient data for a reliable assessment of the status of
the relevant quality element. As a guideline, monitoring should take place at
10
intervals not exceeding those shown … unless greater intervals would be
justified…”(European Commission, 2000). However the role of passive samplers
and stormwater models for monitoring purposes has not been established. The
WFD monitoring requirements are illustrated in Figure 4.
Figure 4: Monitoring strategies as suggested in the water framework directive (European
Commission, 2000). Dotted lines: not specifically mentioned in the WFD.
Surveillance and operational monitoring both target the status of the water
bodies. Therefore the monitoring points are within the respective lakes, rivers or
streams. Long term changes in the chemical status of water bodies with relatively
slow changes in concentrations can be assessed by taking grab samples, since a
grab sample in this case can represent the concentration in the water body during
a time period. However, care should be taken when defining the monitoring point
in surface waters impacted by urban runoff, since stormwater discharges are so
dynamic that one sample per month (as suggested in the WFD) in no reasonable
way can represent the level of pollution if the monitoring point is located close to
a discharge point.
Estimating long term changes in water bodies with fast changes of water flow
and quality, such as streams or rivers, by taking the required amount of grab
samples is not straight forward. For analytes showing a strong daily pattern,
different conclusions can be drawn depending on when the samples are taken,
during day or night, dry or wet weather etc. (Hazelton, 1998). The sampling
strategy should therefore be determined depending on the use pattern for the PPs
in question. Herbicides showing a spring first flush pattern can e.g. be monitored
Surveillance monitoringOperational monitoring
AA-EQS MAC-EQS
Investigative monitoring
Point sources and
stormwater discharges
Grab samples
1 month-1
Volume-proportional
sampling
Lakes, rivers or streams
Monitoring type:
Sampling point:
Sampling method:
11
by sampling more frequently in spring and early summer months and assuming
zero concentrations during the rest of the year (Battaglin and Hay, 1996), on the
other hand Fletcher and Deletic (2007) found that for evaluating the TSS loads in
rivers and streams sampling intervals should be 3 days or less in order to ensure
absolute errors below 10%.
Operational monitoring must relate to the AA-EQSs and MAC-EQSs when
establishing the chemical status of the water bodies. Operational monitoring
strategies aimed at finding annual average concentrations are comparable to
surveillance monitoring strategies for finding long-term changes of the chemical
status of surface waters. In the definition of the MAC-EQS, the requirement is
that none of the samples taken during the monitoring program are allowed to
exceed the MAC-EQS. “However ... Member States may introduce statistical
methods, such as a percentile calculation, to ensure an acceptable level of
confidence and precision for determining compliance with the MAC-EQS”
(European Commission, 2008).
Investigative monitoring is by nature different from the two other types of
monitoring. The aim is not to establish the status but to find out why a good
status is not achieved. The investigative monitoring is therefore not described as
detailed as the two other monitoring strategies. In investigative monitoring it can
be relevant to sample stormwater discharges and point sources. Stormwater
discharges and point sources are much more dynamic in nature than most water
bodies. In a comprehensive study of sampling approaches by Ackerman et al.
(2011), it was concluded that the most efficient sampling method for stormwater
monitoring was volume proportional sampling (using 10 sub-samples per
composite sample), with variable sample pacing to ensure that the entire storm
was captured.
2.2 Monitoring priority pollutants in stormwater runoff As already mentioned, monitoring of stormwater can be used in WFD
investigative monitoring to find out why the receiving waters do not have good
chemical status. However, there are other reasons to monitor stormwater quality.
When looking at the water quality of stormwater runoff it is evident that there is
a need for treatment of stormwater in order to comply with the EQSs in most
receiving waters (especially water bodies which receive a large fraction of urban
stormwater runoff). One approach is to use standard criteria for designing
12
stormwater treatment systems such as ‘best available technology’. However, it is
also expensive to oversize treatment systems or to apply treatment that may not
be necessary at all locations. Monitoring can therefore help in designing
appropriate treatment systems for stormwater runoff. Furthermore, in Denmark,
the water utilities who are responsible for management of stormwater runoff, are
privatized and no longer part of the municipalities. As the responsibility of the
water bodies belong to the municipality, the municipality sets requirements for
the quality of stormwater discharged to the water bodies. It can therefore be
important for the water utilities to be able to document the quality of the water
they discharge to the receiving waters.
13
3 Priority pollutants in stormwater 3.1 Selection of pollutants for monitoring Different approaches are followed when choosing which pollutants to include in
monitoring programs for identification of pollutant concentration levels. Five
examples of approaches are presented here: the scientific approach CHIAT, the
approach used in a research project ESPRIT, the Danish national monitoring
program NOVANA, the US EPA monitoring program for water quality and the
approach used for a screening campaign in Paper I. A minimum data set for
evaluating performance of stormwater treatment systems have also been
identified (Ingvertsen et al., 2011).
The chemical hazard identification and assessment tool (CHIAT) is a scientific
approach to select PPs (Eriksson et al., 2005). The CHIAT method covers source
characterization, exposure identification, hazard identification and assessment as
well as expert judgement. A tool for source characterization has been developed
in the EU project ScorePP. It consists of a database of sources, release factors
and patterns compiled for selected priority substances from the WFD
(www.scorepp.eu) (Lützhøft et al., 2012). However, the source characterization
in the CHIAT method is not limited to regulated pollutant, but aims at a broad
identification of sources and potential pollutants. The hazard identification can be
evaluated using the RICH method (ranking and identification of chemical
hazards) (Baun et al., 2008). In the RICH method, pollutants are ranked based on
their inherent properties. The CHIAT method was used to select a list of
stormwater PPs in the EU Daywater project (Eriksson et al., 2007b).
One of the research projects on stormwater PPs conducted during the same
period as this PhD project is the ESPRIT project, where the aim was to identify,
evaluate and characterize priority substances from the EU WFD in stormwater.
Legal considerations was therefore the (only) criterion for selecting pollutants to
monitor (Bertrand-Krajewski et al., 2008).
In the Danish national monitoring program, substances were selected to ’meet the
international obligations and at the same time provide an overview over the
inputs from the different types of point sources’ (Danish Environmental
Protection Agency, 2000). Here legal considerations are also important; however
the criteria followed for the specific choices of pollutants have not been stated.
14
The approach set out by the US EPA for choosing indicators to include in water
quality monitoring programs is stated this way: “Because limited resources affect
the design of water quality monitoring programs, the State should use a tiered
approach to monitoring that includes a core set of baseline indicators selected to
represent each applicable designated use, plus supplemental indicators selected
according to site-specific or project-specific decision criteria” (US EPA, 2003).
Thus overall considerations as well as local considerations are taken into account,
and the economical aspect is also pointed out.
The criteria used when selecting pollutants for a screening campaign of
pollutants in stormwater runoff are described in Paper I (see Figure 5). Three
criteria were used: legislation, local considerations and practical aspects. Similar
to the ESPRIT project and the Danish national monitoring program, the priority
substances listed in the EQS directive (European Commission, 2008) were used
as a basis for the selection of PPs.
Figure 5: Selection of priority pollutants for monitoring (Paper I). EU priority substances and
priority hazardous substances (bold) are shown in the red rounded shape. Pollutants found in an
earlier risk assessment of a catchment in Copenhagen are shown in the yellow shape. Pollutants
selected in the present study based on practical limitations are shown in the blue shape.
15
As in the approach used by the US EPA, this basic list was supplemented with
local knowledge, in this case a risk assessment performed in the area and
knowledge on the use of pesticides (Eriksson et al., 2007a). However, the
practical side is often the limiting factor when designing monitoring programs.
Therefore considerations of the budget size, which analytes were included in the
same analyses, detection limits and how much sampling volume was needed for
each analysis was also taken into account.
From these monitoring campaigns it is therefore seen, that the important factors
to consider when choosing PPs for monitoring are: scientific knowledge on
pollutants (e.g. inherent properties as used in the RICH method), relevant
legislation, local considerations (sources and earlier findings) and practical issues
(economy, analytical considerations, sampling etc.). The importance of each
factor depends on the local settings and aims. However, when all these factors
are considered, selection can be done in a deliberate manner where the
monitoring program is limited to the important pollutants.
3.2 Sources and occurrence of pollutants in stormwater
The main sources of pollution in stormwater are wet and dry deposition,
activities in the catchment (e.g. traffic, application of pesticides, industrial
activity) as well as release of pollutants from materials in contact with the water
(see e.g. Eriksson et al. (2005)). The sources to the PPs can be activities outside
of the local catchment since dry and wet deposition can transport PPs over short
and long distances. Traffic sources covers wear and tear of tires, asphalt, brakes,
undercoating, and combustion products from exhaust as well as drip losses (e.g.
Sörme and Lagerkvist (2002)). Materials in contact with the water include
asphalt, roof materials, paints, concrete etc. These materials have been shown to
leak pollutants such as biocides (Burkhardt et al., 2011; Schoknecht et al., 2009;
Skarzynska et al., 2007)
A large number of studies have investigated different pollutants in stormwater
runoff. Over 650 organic pollutants and 30 metals and inorganic metals have
been identified as potential stormwater pollutants (Eriksson et al., 2007b; Ledin
et al., 2004). There is a varying level of knowledge about the occurrence of these
pollutants in stormwater. For some parameters such as suspended solids, organic
matter, nutrients and heavy metals, a fair amount of data can be found in the
16
literature (Göbel et al., 2007; Ingvertsen et al., 2011), and they are also included
in the National Stormwater Quality Database, NSQD, from the USA covering
over 3700 storm events (Maestre and Pitt, 2005). For these parameters,
concentrations have been compiled for different catchment types (Maestre and
Pitt, 2005) and surface types (e.g. road and roof types) (Göbel et al., 2007;
Skarzynska et al., 2007).
Information on organic compounds in stormwater is much scarcer than on
suspended solids, organic matter, nutrients and heavy metals. Individual
polyaromatic hydrocarbons (PAHs) have been investigated to some extent, while
herbicides, pesticides and a range of other organic substances have only been
measured in a few studies (Ingvertsen et al., 2011; Ledin et al., 2004; Skarzynska
et al., 2007). Studies which include monitoring of a broad range of pollutants at
the same location or time, e.g. the priority substances regulated through the
WFD, are only recently reported and currently being conducted. This includes
the screening campaign reported in Paper I, which included a broad range of
pollutants at five different locations (4 stormwater discharge sites and 1
combined sewer overflow). In France, large research projects measuring PPs in
atmospheric deposition and stormwater are under way (Bertrand-Krajewski et al.,
2008; Sebastian et al., 2011), and measurements of PPs in surface waters and
stormwater have been reported (Gasperi et al., 2009; Zgheib et al., 2011).
In Table 3 an overview from literature of concentration ranges of PPs in
stormwater is shown (basic parameters such as TSS, organic matter and nutrients
are not included). Different levels of smoothing of the data have been used in the
different reviews and studies: in some studies concentration ranges in single
samples are averaged into EMCs for each event, and in other studies these are
averaged to SMCs. For comparison, the MAC-EQS, and where this is not defined
the AA-EQS are listed as well. Note that the EQS values for the heavy metals
apply to measurements of filtered samples, while the literature listed here reports
total concentrations (dissolved concentrations can vary from 90% to <10%
depending on the heavy metal of interest (Ingvertsen et al., 2011)).
As mentioned earlier, the larger (review) studies by Maestre and Pitt (2005) and
Göbel et al. (2007) include mainly information about heavy metals. As the study
by Ledin et al. (2004) includes concentrations of samples rather than EMCs or
SMCs, the range is broader than the other review studies, and extreme
17
concentrations are also included which are not representative of the general state
of stormwater runoff. However it is evident that stormwater from some sites
would require much treatment in order to comply with the guidelines in the
recipients. A proposal for an update of the EQS directive (European
Commission, 2011) lowers the EQSs for PAHs, brominated diphenylethers, lead
and nickel compared to the existing EQSs (EC 2008) used in this PhD thesis (e.g.
AA-EQS and MAC-EQS for fluoranthene is lowered from 0.1 and 1 µg/L to
0.0063 and 0.12 µg/L respectively), meaning that in the future it is anticipated to
be even harder to comply with EQSs in surface waters.
Which pollutants are problematic in stormwater runoff depends on the actual
location. However some pollutants are more general stormwater pollutants than
others. The heavy metals have for a long time been in focus in stormwater runoff.
Very high concentrations have been found in stormwater runoff compared with
the EQSs for surface waters (Table 3). However, the concentrations measured in
the screening study (Paper I) and in the stormwater pond in Albertslund (Paper
IV and V), indicate that the heavy metals Cu and Zn, which are not strongly
sorbed to particulate matter in stormwater (sorption of 10-95% have been found
for Zn and 35-90% for Cu (Ingvertsen et al., 2011)), are most problematic in
stormwater discharges at these sites.
PAHs in stormwater runoff are seen to exceed the EQSs in all studies. As PAHs
are related to traffic and fireplaces these compounds are more generally found in
stormwater runoff than other point source related compounds. Removal of PAHs
in stormwater treatment facilities is therefore important.
Pesticides can be regulated through source control much easier than the PAHs.
Many pesticides are regulated in use. Pesticides which are banned are often only
found in low concentrations in stormwater runoff. However, for many of the
pesticides which are in use, and therefore found in higher concentrations, EQSs
have not yet been established. In order to evaluate the effect of pesticides in
stormwater runoff, toxicity data of the locally used pesticides have to be
consulted.
Table 3: Concentration of selected priority pollutants in stormwater. The table is not exhaustive for constituents and studies.
Pollutant µg/l
Review studies Broad studies MAC-EQS/(AA-EQS)f
(Maestre and Pitt, 2005)a
(Göbel et al., 2007)b
(Ledin et al., 2004)c
(Zgheib et al., 2011)d
Paper Ie
WFD
Heavy metals
Cd 0.2-13 <0.1-700 n.d. 0.11-0.63 0.45-1.5g,h
Cr 2-50 <0.5-4200 10-45 0.41-41.2 (3.4)g,i
Cu 0.6-1360 6-3416 <0.5-6800 50-220 22-255 (12)g,i
Ni 1-120 2-70 5-580 n.d. 0.93-40.5 (20)g,i
Pb 0.2-1200 2-525 <0.5-2764 25-129 9.8-72.4 (7.2)g
Zn 2-22500 15-4880 0-38061 130-520 74-244 (7.8)g,i
PAHs
Naphthalene 0.006-49 0.088-0.175 <0.01-0.72 (2.4)
Acenaphtylene <0.05-0.96 0.027-0.126 <0.01-0.039
Acenaphthene 0.002-0.97 0.013-0.044 <0.01
Fluorene 0.001-74 0.019-0.106 <0.01-0.028
Phenanthrene <0.01-1420 0.090.0.712 0.017-0.29
Anthracene <0.0001-147 0.016-0.096 0.012-0.084 0.4
Fluoranthene 0.009-1958 0.098-0.832 0.025-0.55 1
Pyrene 0.0001-120 0.100-1.223 0.034-0.56
Benzo(a)anthracene 0.0003-54 0.037-0.298 <0.01-0.21
Chrysen/triphenylene <0.01-2271 0.088-0.655 <0.01-0.38
Benzo(b+j+k)fluoranthene <0.01-0.49 0.100-0.876 <0.01-1.0 (0.03)
Benzo(a)pyrene 0.00015-300 0.041-0.315 <0.01-0.31 0.1
Benzo(g,h,i)perylene <0.01-710 0.071-0.569 <0.01-0.47 (0.002)
Indeno(1,2,3-cd)pyrene <0.01-1080 0.053-0.354 <0.01-0.39
Dibenzo(a,h)anthracene <0.01-83 0.021-0.096 <0.01
Total PAH 0.24-17 <0.011-178 0.088-4.38
19
Pesticides Glyphosate <0.05-1.92 0.043-1.2
AMPA 0.479-0.731 0.06-0.33
Diuron 0.25-238.4 0.394-0.647 <0.01-0.055 1.8
Isoproturon <0.05-0.079 0.004-0.082 <0.01-0.044 1.0
Terbutylazine 0.017-0.16 <0.01
MCPA 0.009-0.13 <0.01-0.018
Monobutyltin 0.091-0.120 0.035-0.048
Dibutyltin 0.074-0.093 0.008-0.009
TBT 0.050-0.078 <0.004 0.0015
Other organic substances
Benzene 0.017-13 n.d. <2 50
Nonylphenol <0.04-23 1.595-9.17 <0.1-0.43 2.0
Octylphenolethoxylates <0.5-12 (0.1)
DEHP 3.0-44 15.3-60.9 2.9-8.5 (1.3)
Trichloroethylene 0.036-7 n.d. <0.02 (10)
Tetrachloroethylene 0.058-25 n.d. <0.02 (10)
Chloroform <0.1-12 n.d. 0.02 (2.5) aThe review reports range of site mean concentrations (SMCs) with a minimum number of events for each site of 3.
bThe review reports event mean
concentrations (EMCs). cThe review reports sample concentrations (all kind of samples).
dThe study reports EMCs.
eThe study reports grab samples and
precipitation proportional samples. fmaximum allowable concentration or annual average concentration from the water framework directive
gfiltered
samples, hdepending on the hardness of the water,
iDanish EPA guidelines, n.d.: not detected. Blank cell = not included in the study.
20
There are many other organic pollutants which could be problematic at specific
sites. However, DEHP is generally found in stormwater runoff. This is connected
to the wide use of this plasticizer in plastic products which are exposed to
weathering.
The main pollutants were therefore found to be the heavy metals Cu and Zn, the
PAHs, pesticides in use (here Glyophosate was found) and the plasticizer DEHP.
21
4 Partitioning of priority pollutants in stormwater
Sorption of PPs to particulate matter and dissolved organic carbon (DOC) in
stormwater is important for the toxicity, degradation and transport of PPs, and for
removal in stormwater treatment systems.
4.1 General principles Partitioning or speciation is the distribution of a compound between its chemical
forms e.g. freely dissolved, ionic, complex-bound, particle-bound, bound to
colloids and taken up in biota. The term speciation is often used when talking
about metals, and the term partitioning when referring to organic compounds.
The distribution is characterized by equilibrium constants, partition ratios or
accumulation factors and depends on the compound as well as on the constituents
in the system; ligands, organic substances, particles etc. (Figure 6). The
speciation of metals and hydrophobic organic compounds are driven by different
mechanisms. Freely dissolved metals are on ionic form and interact with other
ions, macromolecules or colloids by forming complexes and with suspended
organic and inorganic particles by sorption-desorption processes. Hydrophilic
compounds can be either on ionic or non-ionic form, and can distribute between
the different forms in a manner similar to metals. Hydrophobic organic
compounds are not on the ionic form and do not form complexes but distribute
among the organic phases of the system mainly by absorption and adsorption (if
the compounds can be ionized, the ions can also form complexes in a manner
similar to metals).
Figure 6: Speciation of metals (a) and hydrophobic organic compounds (b). Dashed arrows
indicate that this uptake route is only relevant for some type of biota.
Metal ion
Sediment
Suspendedsolids.
....
.
.. .... ..
Dissolved organicmatter
Living organisms
Inorganicligands..
....
a)
Freely dissolved
Sediment
Suspendedsolids.
....
.
.. .... ..
Dissolved organicmatter
Living organismsb)
22
Kinetics of complexation and sorption are very important for speciation. Two
different forms can be distinguished – labile and inert forms. Labile species are
the forms which have relatively fast complexation or partitioning kinetics
compared to the time-frame of the system, whereas inert forms have slow
kinetics (species with kinetics in the same range as the time-frame of the system
are called non-labile). Labile species are therefore in rapid equilibrium with each
other whereas inert species do not obtain equilibrium and do not contribute
noticeably to partitioning processes. As the time-scale of the system defines the
frame for the equilibrium, species can be considered labile in one system and
inert in another system.
Uptake of PPs in organisms and in many passive samplers is constricted to the
freely dissolved and labile species. Exceptions from this and discussion of the
uptake mechanisms of metals and organics in aquatic organisms is out of the
scope of this thesis, but is discussed in more depth in e.g. Worms et al. (2006).
4.2 Methods for speciation and partitioning measurements
For characterization of speciation in aqueous samples it is necessary to be able to
analytically distinguish the complexed and/or sorbed species from the freely
dissolved compound. For metals, van Leeuwen et al. (2005) go into depth with
the concept of dynamic speciation and the available sensors for metal speciation
measurements. Since sorption is more important for HOCs than for hydrophilic
compounds and most of the organic pollutants included in the WFD are
hydrophobic, partition measurement methods will here be discussed with a focus
on the HOCs.
Speciation of HOCs is theoretically not as complicated as for heavy metals
because inorganic complexes can be neglected and because partitioning
processes are well defined. However sorption processes with very slow kinetics
(Jonker et al., 2005) or ‘trapping’ of compounds in some types of particles has
been suggested for some HOCs making them unable to participate in the
partitioning processes (Jonker and Koelmans, 2002).
The most ‘simple’ way to determine partition ratios to particulate material is to
separate the water phase and particulates, either by centrifugation and/or
filtration (0.45µm is the standard, however sometimes other filter sizes are used),
23
and measure the concentration of the compound in both the water phase and in
the particles. This method is however only reliable if partitioning to dissolved
organic matter (DOM) is negligible, otherwise it will underestimate the partition
ratio (Lee et al., 2003). Furthermore the method depends on the strength of the
solvent used for extraction. In cases where some of the compound is trapped
inside particles, full extraction by strong solvents will give information about the
actual distribution of HOCs in the sample, however not on the partition ratio (as
some of the measured compound is not in equilibrium).
Multiple different methods have been developed in order to separate the
compounds sorbed to DOM from the freely dissolved compound. In the
complexation-flocculation method separation of the complex-bound compound
from the freely dissolved compound is obtained by flocculation (Laor and
Rebhun, 1997), whereas in the reverse phase separation method the freely
dissolved form is adsorbed on a C18-column assuming that the DOM-bound
compound passes the column (Landrum et al., 1984). However, for both methods
insufficient separation can lead to bias on partition ratios.
Other techniques which are more appropriate for measuring partition ratios
include the method developed during this PhD called passive dosing (Paper II),
the solid phase micro extraction method (SPME) (Arthur and Pawliszyn, 1990)
and the solid phase dosing and sampling method (Ter Laak et al., 2005). These
methods do not include phase separation of the sample. Fluorescence quenching
(Gauthier et al., 1986) is another method for measuring partition ratios without
the need for phase separation. However, fluorescence quenching has the draw-
back, that it is restricted to compounds with low hydrophobicity (e.g. four ring
PAHs or smaller) because the fluorescence of the freely dissolved concentration
has to be measureable (Raber et al., 1998). Furthermore, the assumption that
partitioning to all types of organic matter results in static quenching of the
fluorescence is questionable (e.g. Puchalski et al. (1992)).
Both the passive dosing method and the solid phase dosing and sampling method
employ analytes dosed to the sample and not the native analytes in the sample.
The advantage of this is that the partition ratio will then be measureable even in
situations where the native analytes are below detection. The advantage of SPME
is that the concentration of the native analyte is determined at the same time as
the speciation and this additional information can show the relevant
24
concentrations on each form. However, the passive dosing method avoids some
of the analytical problems which can be encountered when using SPME or solid
phase dosing and sampling, the most crucial being avoiding a complete mass
balance assumption (in passive dosing, analytes sorbed to the glass vial will be
replenished by the dosing from the PDMS), but also avoiding fouling problems
and ensuring equilibrium in a practical time period (Paper II). Furthermore,
passive dosing has been shown to be simple, precise and robust.
4.3 Partitioning of PAHs Most of the literature investigations of the partitioning of PAHs to particulate
matter have been conducted using lake, river and saline sediments, although
some investigations of partitioning to soil, river and estuarine suspended solids
(SS), stormwater particles and different types of soot and biofilms have also been
reported (see Table 4).
Partitioning of PAHs to dissolved matter has been investigated in sediment and
soil samples. Table 4 shows a selection of partition ratios for fluoranthene found
in the literature. The reported partition ratios cover three orders of magnitude.
Similar sized ranges can be found for other PAHs (Hawthorne et al., 2006;
Kumata et al., 2000). This level of variation cannot be explained by uncertainty
in the partition measurements. It is therefore evident that the nature of the
particulate matter has a high influence on the partitioning. Studies have shown
that the amount and degree of aromaticity of the organic matter influences
partition ratios (Gauthier et al., 1987; Neale et al., 2011) and high partition ratios
were found to soot particles (Jonker and Koelmans, 2002; Zhou et al., 1999).
Partition in stormwater runoff has not been studied much. Kumata et al. (2000)
measured partition in runoff from a highly trafficked road in Tokyo, and in
Paper III passive dosing measurements of partition in runoff from an industrial
and residential area (Albertslund) is reported. Furthermore, partition ratios have
been measured in samples from a stormwater pond in Lyon (not published but
shown in Figure 7 and Figure 8).
In the passive dosing method, partition ratios can be found by plotting the
enhanced capacity of the sample for the substance caused by TSS, ETSS, (this
property is explained and defined in Paper III) against TSS or particulate
organic matter, POM (POM can be used if it is assumed to dominate the sorption
25
compared to the TSS). The very high correlation between the POM and ETSS in
the samples from Albertslund, indicate that the POM is mainly responsible for
partitioning (Paper III). At this site POM comprises the main fraction of the
particulate matter (24-100%, see Figure 9). Stormwater from the pond in Lyon
showed much lower organic matter content (2-7%). However, the same overall
sorption to TSS was seen in the samples from Lyon compared to the samples
from Albertslund (Figure 8).
Table 4: A selection of partition ratios for fluoranthene found in literature. Extended version of
table 3 in Paper III.
Log K
(L/kg)
Study Dosed
PAH 3
Phase
separation
Suspended solids & dissolved matter1
Stormwater SS
4.58 Paper III Yes No4
Stormwater POM
5.18 Paper III Yes No4
Traffic soot
6.96 (Jonker and Koelmans, 2002)
No No5
6.39 (Jonker and Koelmans, 2002) Yes No5
Wastewater SS
4.3 Paper II Yes No4
Estuary SS 3.8-5.5 (Zhou et al., 1999) No Yes
Soil DOM
4.14, 4.18 (Raber et al., 1998) Yes No7
Organic carbon2
Estuary suspended OC
6.51 (Fernandes et al., 1997)
No Yes
4.6 (Tremblay et al., 2005) Yes Yes
5.6-7.2 (Zhou et al., 1999) No Yes
Highway runoff OC
4.8-7.2 (Kumata et al., 2000) No Yes
Sediment OC
4.32-7.5 (Hawthorne et al., 2006)
No Yes
5.4 (Luers and tenHulscher, 1996) Yes No6
6.08-6.66 (McGroddy and Farrington, 1995) No Yes
4.75 (Persson et al., 2005) No Yes
4.89;5.32 (de Maagd et al., 1998) Yes Yes
Soil OC
4.65-4.82 (He et al., 1995)
Yes Yes
Biofilm OC 4.5 (Wicke et al., 2007) Yes Yes
Sediment DOC
5.18 (Luers and tenHulscher, 1996) Yes No6
4.75; 5.41 (Haftka et al., 2010) No No
8
4.2-5.0 (Brannon et al., 1995) Yes Yes
AHA
5.26 Paper II Yes No4
1Partition ratios found on a total weight basis.
2Partition ratios found on an organic carbon basis.
3Partition ratio calculated based on dosed PAH,
4measured by passive dosing,
5measured by
poluoxymethylene-SPE partitioning method, 6measured by gas purge,
7measured by
fluorescence quenching, 8measured by SPME. SS=suspended solids, POM=particulate organic
matter, DOM=dissolved organic matter, OC=organic carbon, AHA=Aldrich humic acid.
26
Figure 7: Enhanced capacity caused by suspended solids, ETSS, plotted against the concentration
of particulate organic matter (POM). Stormwater samples from Albertslund (green, red and blue
symbols) and Lyon (orange crosses). Revised from Figure 4 in Paper III.
Figure 8: Enhanced capacity caused by suspended solids, ETSS, plotted against the concentration
of TSS. Stormwater samples from Albetslund (black dots) and Lyon (orange crosses).
Regression is performed on all samples, and the partition ratio and 95 % confidence interval on
the partition ratio is given. Error bars show standard deviation on each sample (n=3).
Figure 9: Relationship between particulate organic matter, POM, and total suspended solids,
TSS, in samples from the inlet and outlet of a retention pond in Albertslund (black dots) and
Lyon (orange crosses).
0 50 100 1500
10
20
[POM] (mg/L)
ET
SS
0 200 400 600 8000
10
20
30 KTSS = 38800 ± 1000 L/kg
R2 = 0.92
[TSS] (mg/L)
ET
SS
0 200 400 600 8000
50
100
150
200
Lyon
Albertslund
TSS (mg/L)
PO
M (
mg/L
)
27
This can be explained if the particulate organic matter is either very different
from the organic matter in samples from Albertslund or if the sorption to
inorganic matter play an important role in the sorption of fluoranthene to TSS in
stormwater with low POM concentrations.
Whereas Kumata et al. (2000) found highly varying partition in stormwater
samples from the same site, this was not the case in our investigations of data
from Albertslund and Lyon (Figure 7 and Figure 8). This is probably due to the
different measurement method used for the two studies. Kumata et al. (2000)
separated the two phases and extracted the PAHs from the particles by
dichloromethane. If some of the PAHs on the traffic particulates are not available
for partitioning, they will be included in the partition ratio thus obtained.
However, the passive dosing method, which was used in Paper III, measures
equilibrium partitioning. These findings could indicate that whereas the
equilibrium partitioning might be relatively constant for the same site over time,
the amount of PAHs which are strongly bound can vary. However the varying
partition found by Kumata et al. (2000) could also be explained if the type of
organic material in the stormwater varied over time at the site they investigated.
As shortly suggested in Paper III, the partition ratio measured for fluoranthene
by passive dosing can be used to estimate partition ratios for the other PAHs at
the same site. The assumption that log Kow and log KPOM are linearly related with
a slope of approximately one (Witt et al., 2010), leads to the relationship:
Log KPOM,X = log Kow,X - log Kow,fluoranthene + log KPOM,fluoranthene
Solving equation (4), (10) and (11) from Paper III for the free fraction, ff, leads
to the relationship:
1
1
DOCKPOMKff
DOCPOM
The estimated free fractions of five other PAHs in the stormwater samples from
Albertslund containing the highest and lowest amount of POM, are shown in
Table 5. The estimated ffs illustrate the large difference in environmental
behavior of the PAHs, where naphthalene is mainly freely dissolved in
stormwater at this site whereas the 5 and 6 ring PAHs (bottom three in Table 5)
28
are mainly sorbed to organic matter in the stormwater. The ffs also indicates the
fraction of the pollutants which are available for uptake in organisms.
Based on the present data, it is suggested for modeling purposes to use a log KTSS
of 4.59 for partitioning of fluoranthene to TSS in stormwater if no measurements
are done locally.
Table 5: Free fraction of PAHs estimated for the stormwater runoff sample from Albertslund
with highest and lowest concentration of particulate organic matter, POM, (The samples can be
seen on Figure 7 as the dots furthest to the right and left, respectively).
ff at max POM ff at min POM
Napthalene 0.8 1
Anthracene 0.1 0.8
Fluoranthene 0.04 0.5
Benzo(k)fluoranthene 0.006 0.1
Benzo(a)pyrene 0.005 0.1
Benzo(ghi)perylene 0.004 0.1
29
5 Sampling priority pollutants in stormwater
In this PhD study focus has been on sampling for chemical analysis. In the
following section uncertainties, advantages and disadvantages of the traditional
sampling methods are discussed and in Section 5.2, passive samplers are
presented and discussed.
5.1 Traditional methods The three most common stormwater sampling methods are grab, time- and
volume proportional sampling. Flow proportional sampling is not as common
(sampling a water volume proportional to the flow at constant time intervals as
opposed to the volume proportional sampling where a fixed water volume is
sampled at variable time intervals proportional to the flow). Special techniques
for sampling runoff water close to the source, e.g. roof runoff from down-pipes
and road runoff at road level and small ditches, is presented in Skarzynska et al.
(2007). In Table 6 advantages and disadvantages for the most common methods
are listed.
Table 6: Advantages and disadvantages of grab, time and volume proportional sampling.
Advantages Disadvantages
Grab sampling Low cost
Special handling can be
incorporated
No volume restrictions
Represents only one point in time
Logistic difficulties (sampling during
holidays and at night etc.)
Time proportional
sampling
No need to be at site during
rain
More representative than grab
sampling
Rainfall forecast important (start and
duration)
Two field trips necessary (start of
equipment and retrieval of samples)
Bias on load estimates at varying flows
Heavy equipment/permanent
installation
Volume
proportional
sampling
No need to be at site during
rain
More representative than time
proportional for varying flows
Rainfall forecast important (start,
duration and amount)
Two field trips necessary (start of
equipment and retrieval of samples)
Heavy equipment/permanent
installation
30
Time and volume proportional samples are taken either as samples drawn at a
single point in time, or as one or more composite samples consisting of a number
of sub-samples. It should be noted that representativeness of the time and volume
proportional sampling methods depends on the temporal resolution between
samples or sub-samples. Therefore when reporting sampling programs it is
important to describe the time/volume between sub-samples as well as how many
sub-samples are used for composite samples. Detailed description of the
sampling is also important because there is not agreement in the scientific
community on the terminology used e.g. volume proportional sampling (Ort et
al., 2006) is also called flow weighted sampling (Maestre and Pitt, 2005; US
EPA, 1992), flow proportional sampling (Lindblom et al., 2011), equal discharge
sampling (Ma et al., 2009) and volume paced composite/micro sampling
(Ackerman et al., 2011). In order to avoid misunderstandings it should therefore
be evident in the description how the sampling was done.
Fletcher and Deletic (2007) found that a monitoring program based on single
grab samples taken 1 hour after the rain start resulted in good estimates on annual
loadings (even though EMC estimates were not always accurate). However, if the
EMC should be accurately described, it has been shown that more than 30
samples has to be taken to estimate EMCs within 20% average error (Ma et al.,
2009). It has also been shown that volume proportional sampling is more
accurate than time proportional sampling (Leecaster et al., 2002; Ma et al., 2009)
and that rainfall depth proportional sampling had accuracies between the two
other methods (Ma et al., 2009). Ackerman et al. (2011) conclude that
pollutograph sampling (by Ackerman et al. used to denote that a series of
samples are taken during the rain event and a sub-set of the samples are analyzed
based on the flow data retrieved) has the highest accuracy (how close the
measured value is to the true value) and precision (deviation of repeated
measurements) for estimating EMCs. However, an adaptive targeted sampling,
where the pacing is adjusted based on the anticipated size of rain events can have
equally high performance as nearly 100% of the samplings would have EMCs
within 10% of the true EMC if the actual rain event size is between 50-100% of
the anticipated event size (Ackerman et al., 2011). Sampling campaigns should
therefore be designed taking into account the needed accuracy and precision of
the results.
31
The uncertainties in estimating EMCs when using automated samplers originate
from variations in space as well as in time. Vertical distribution of suspended
solids (and associated pollutants) can be observed in flow channels with larger
particles being transported near the bottom, and for rivers the horizontal
distribution can be important as well (Horowitz, 2008). Roseen et al. (2011)
found that sediment EMCs were well represented using an automated sampler
compared to measurements where they collected the total runoff volume of the
event for analysis. However larger particles may fall out of suspension in the
sampling hose if the suction velocity is lower than the flow velocity at the site.
Suggestions on how to optimize the intake for autosamplers have been published
by Gettel et al. (2011) who designed a multiple tube intake fitted with a wing to
provide lift.
5.2 Passive sampling There exist two types of passive samplers. Diffusion based passive samplers and
flow-through passive samplers. Most of the literature published so far, and most
of the passive samplers developed are diffusion based. The following section
describes these passive samplers, and is based on work published in (Pettersson
et al., 2010). However, when monitoring stormwater discharges over periods of
weeks or months to evaluate average concentrations or loads, time-dependent
sampling (as obtained by diffusion based passive samplers) is not suitable
because the dry periods between rain events would be included in the
measurements. This would bias the measurements of the discharged
concentrations. In this PhD project, it was therefore decided to test a flow-
through passive sampler, SorbiCell, which can be installed so that sampling
depends on the water velocity at the site. This passive sampler type is described
further in section 5.2.2, however, an important difference between the two types
of samplers is that diffusion based samplers by nature only samples the freely
dissolved and labile species, whereas flow-through samplers can collect any
fraction which is able to pass through the inlet filter.
5.2.1 Diffusion based passive samplers
The principle behind diffusion based passive sampling is that a receiving phase,
which has a high affinity for the compounds of interest, is placed in the medium
to be sampled; e.g. water. The sampler can be with or without a membrane
separating the receiving phase and the sampled phase. The passive sampler can
then, based on chemical or physical properties of the membrane, receiving phase
32
and analytes, selectively accumulate chemical substances by diffusion from the
water to the sampler. During the last decades, a lot of work has been put into the
development of these passive samplers, and a range of reviews have been
published covering the area (Namiesnik et al., 2005; Seethapathy et al., 2008;
Stuer-Lauridsen, 2005; Vrana et al., 2005). Also, comparative studies including
many different passive samplers are now being reported (Allan et al., 2009; Allan
et al., 2010).
In Table 7 advantages and disadvantages of diffusion based passive sampling are
listed. The main advantage is the time-integrative measurements obtained. There
are both positive and negative practical aspects of passive sampling (which
aspects are most important will depend on the sampling campaign considered),
and the fact that diffusion based passive samplers only sample the freely
dissolved and labile species can be either an advantage or disadvantage
depending on the aim of the study.
Table 7: Advantages and disadvantages of diffusion based passive sampling.
Advantages Disadvantages
Passive sampling Continuous sampling
Low detection limits can be
obtained through high pre-
concentration at site
Reduced need for clean-up of
the extract
Freely dissolved and labile
species are sampled
simulating the species
available for uptake in biota.
Transport of large water
volumes are not necessary
Risk of contamination for some of the
samplers which are effective air
samplers as well.
Freely dissolved and labile species are
sampled, however total concentrations
are regulated
Vandalism/loss of samplers
Fouling
There are two different uptake mechanisms in diffusion based passive samplers:
absorption and adsorption (see Table 8). Passive samplers targeting HOCs
generally work by absorption (partition based samplers) and passive samplers
targeting polar organic compounds and heavy metals work by adsorption (surface
bonding).
33
Table 8: Overview of the most common passive samplers.
Analytes Hydrophobic organic
compounds
Polar organic
compounds
Metals
Accumulation
mechanism
Absorption Adsorption
Uptake regime Kinetic Equilibrium Kinetic
Samplers SPMD LDPE
Silicone rubber Chemcatcher
SPME Stir bar sorptive
extraction
Chemcatcher POCIS
Chemcatcher DGT
Calibration Sampling rate Rs can be found by PRCs for each
deployment
Partition ratio Ksw found by lab
calibration or from log Kow
relations
Sampling rate, Rs, are estimated in lab/field calibrations. They vary
depending on the analyte, velocity around the sampler, temperature and
biofouling.
Uptake in partition based passive samplers can be described by the partitioning
theory:
sSW
s
SWwsVK
tRKCC exp1
where Cs is the concentration in the sampler, Cw is the concentration in the water,
Ksw is the partition ratio between the sampler and the water, Vs is the volume of
the sampler, Rs is the sampling rate and t is the sampling time (Huckins et al.,
1993). Depending on the configuration of the sampler and the deployment time,
partition based passive samplers can operate as either kinetic or equilibrium
samplers (see Figure 10).
Figure 10: Uptake in kinetic and equilibrium partitioning passive samplers (Allan et al., 2006b).
00
Kineticsamplers
Equilibriumsamplers
Time
Concentr
ation in s
am
ple
r
34
Equilibrium passive samplers are deployed for a period sufficient to obtain near-
equilibrium between the sampler and water. Requirements for equilibrium
sampling are that the response time of the sampler should be shorter than
fluctuations in analyte concentration at the site and that the amount of the analyte
taken up by the sampler should be much smaller than the amount of the analyte
in the sampled water in order to avoid depletion of the sampled water. This
method is equivalent to grab sampling, but it has the advantages that
transportation of large sample volumes is avoided, cleaning of extracts can be
minimized and lower detection limits often are obtained as the extraction is
performed in situ. An example of this is the stir bar sorptive extraction technique
(Baltussen et al., 1999). The same principle is also used in laboratory extraction
procedures such as solid phase microextraction (SPME) (Arthur and Pawliszyn,
1990).
Kinetic passive samplers operate in the kinetic uptake regime during the whole
deployment. The uptake at each point in time is dependent on the (varying)
concentration at the site. Therefore this type of sampler operates as a time-
integrative sampler. Many different designs of kinetic passive samplers have
been described in the literature (see Figure 11 for examples). Kinetic samplers,
which have been used quantitatively in the field and which are commercially
available, include (Table 8): semi permeable membrane devices (SPMDs)
(Huckins et al., 1990) used for hydrophobic substances, polar organic chemical
integrative samplers (POCISs) (Alvarez et al., 2004) used for polar organic
substances, diffuse gradient in thin films (DGTs) (Davison and Zhang, 1994;
Zhang and Davison, 1995) used to sample metals and Chemcatcher (Kingston et
al., 2000) which have different versions for sampling of hydrophobic and polar
organic substances as well as metals.
Figure 11: SPMD (left), POCIS (middle) and Chemcatcher (right). Reprinted from Seethapathy
et al. (Seethapathy et al., 2008) with permission from Elsevier.
Membrane
Receivingphase
Poly-propylenebody
Polypropylenesupport disk
Polypropylenescrew lid
SorbentHole for screw
Stainlesssteelwashers
Polyethersulfonemembrane
35
Silicone rubber and low density polyethylene (LDPE) rods or strips (Booij et al.,
2002; Müller et al., 2001) are also used widely (Allan et al., 2009; Allan et al.,
2010; Mills et al., 2011). They are not possible to purchase as a passive sampler
product, but the materials are easy to prepare.
When calculating (average) concentrations from the sampler measurements, the
partition ratio between the analyte and the sorbent in the sampler should be
known for equilibrium samplers, and the sampling rate should be known for
kinetic samplers (Table 8). Partition ratios can be estimated based on empirical
relationships with log Kow (Greenwood et al., 2007; Huckins et al., 2006).
The sampling rate, Rs, depends on factors such as sampler configuration,
temperature, biofouling and velocity around the sampler as well as on the
specific analyte. In order to find the appropriate sampling rate for each specific
deployment, performance reference compounds (PRCs) are used in absorption
based passive samplers. PRCs are compounds which do not occur in the
environment, but are added to the sampler before deployment. The elimination
rate of the PRC from the sampler is then used to compensate for the influence of
the environmental factors at each deployment. The compound specific effect on
the sampling rate can be estimated based on relationships with log Kow (see e.g.
Huckins et al. (2006) and Vrana et al. (2007)), and the sampling rate varies
linearly with the surface area of the sampler (Alvarez et al., 2004; Huckins et al.,
2006; Zhang et al., 2008).
Unlike the absorption based passive samplers, no theoretical models based on
chemical properties are available for predicting the uptake of analytes into
adsorption based passive samplers (Mills et al., 2011). For passive samplers
based on adsorption, sampling rates for each analyte and set of environmental
conditions therefore has to be found from calibration experiments. Efforts are
being made to get around this problem by designing the passive sampler so that
the membrane is rate limiting for the uptake and elimination of analytes
(Mazzella et al., 2007; Persson et al., 2001; Tran et al., 2007).
5.2.2 Flow-through passive samplers
The flow-through passive sampler, SorbiCell, consists of a cartridge filled with a
sorbent and a tracer salt (Figure 12) (de Jonge and Rothenberg, 2005).
36
Figure 12: The flow-through passive sampler, SorbiCell.
The uptake mechanism of the flow-through sampler is different from the other
passive samplers described above, because uptake of the analytes is not based on
diffusion of the analyte towards the passive sampler, but rather on uptake from
water passing through the sampler. The original method used to induce water
flow through the sampler was by mounting the sampler on a submerged reservoir
with atmospheric pressure inside (Rozemeijer et al., 2010). The pressure gradient
from the water column above the sampler resulted in a slow flow of water
through the sampler.
During this PhD project another installation method was tested (Paper IV). This
method consists of placing the sampler directly in the water letting the
momentum from the water velocity induce the flow through the sampler. This
installation method results in a varying sampling rate depending on the velocity.
The sample would ideally represent the load of the analyte passing the sampling
point. However the necessary assumptions in order to obtain this is that (a) the
water velocity is linearly proportional to the water flow at the site, (b) there is no
uptake in the sampler when there is no flow in the system, (c) the water velocity
at the site is proportional to water flow through the sampler and not changing
over time and (d) the water composition of the water passing the sampler can
represent the composition of the water throughout the cross-section of the
channel/ditch (spatial representativeness).
In Paper IV the first three assumptions are discussed. It is argued that (a) even
though the relationship between the water velocity and flow changes over time,
this is not crucial for the sampling method, (b) the maximum uptake of analytes
in the sampler during dry weather is < 50% of the uptake during rain, and most
likely much less, (c) proportionality between the velocity at the site and flow
through the sampler is reasonable but not exactly good (Kronvang et al., 2010).
Sorbent Tracer salt
37
The last assumption (d) is relevant for all sampling methods and has been
discussed above (Section 5.1). Furthermore, based on literature findings in Paper
IV it is concluded that sampling of the particulate fraction smaller than e.g. 150
µm would comprise the majority of the pollutants in the stormwater. However
the actual pore size of the spheriglass inlet filter used in the sampler is not
known.
Flow-through passive samplers targeting heavy metals and samplers targeting
PAHs were tested during this PhD project. Results from the PAH measurements
(Table 9) show that the passive sampler has high detection limits (LODs) for the
PAHs resulting in detection of very few PAHs. For some of the PAHs detection
limits were higher than the AA-EQSs. Phenanthrene was the only PAH detected
in both the replicates during the first installation period, however the precision
was poor with replicate concentrations of 0.93 and 0.05 µg/L.
Table 9. Dublicate (period 1) and triplicate (period 2) PAH concentrations measured by flow-
through passive samplers.
AA-EQS
Period 1
Period 2
mL sampled
90 220
320 310 370
Naphthalene 2.4
<0.29 <0.12
<0.08 <0.08 <0.07
Acenaphthylene
<0.06 <0.02
0.06 <0.03 <0.03
Acenaphthene
<0.06 <0.02
<0.02 <0.02 <0.01
Fluorene
<0.29 <0.12
0.03 <0.02 <0.01
Phenanthrene
0.93 0.05
<0.16 <0.16 <0.14
Anthracene 0.1
<0.06 <0.02
<0.16 <0.16 <0.14
Fluoranthene 0.1
<0.12 <0.05
<0.08 <0.08 <0.07
Pyrene
<0.12 <0.05
<0.03 <0.03 <0.03
Benzo(a)anthracene
<0.12 <0.05
0.06 <0.02 <0.01
Chrysene
<0.12 <0.05
<0.02 <0.02 <0.01
Benzo(b)fluoranthene 0.03
1
<0.12 <0.05
0.04 <0.03 <0.03
Benzo(k)fluoranthene
<0.06 <0.02
<0.03 <0.03 <0.03
Benzo(a)pyrene 0.05
<0.06 <0.02
0.1 <0.03 <0.03
Benzo(ghi)perylene 0.002
2
<0.12 <0.05
0.13 <0.03 <0.03
Indeno(1,2,3-cd)pyrene
<0.12 <0.05
<0.03 <0.03 <0.03
Dibenzo(ah)anthracene
<0.12 <0.05
<0.03 <0.03 <0.03
sum 16 PAH EPA
<6.99 <2.76
<1.9 <1.92 <1.64 1Sum of Benzo(b)fluoranthene and Benzo(k)fluoranthene,
2Sum of Indeno(1,2,3-cd)pyrene and
Benzo(ghi)perylene.
38
For the second installation period Benzo(a)anthracene, Benzo(a)pyrene,
Benzo(ghi)perylene, acenaphthalene, fluorene, Benzo(b)fluoranthene were
detected, but only in one of the 3 replicates.
It was also seen (Table 9 top) that there was a large variation in the amount of
water passing through the samplers in period 1, whereas during period 2 the
sampled amounts of water were not as varying as for period 1. Re-sizing the
flow-through passive sampler for PAHs in order to allow larger volumes of water
to pass through the sampler, is essential to obtain the necessary low detection
limits.
The results of the heavy metal passive samplers, presented in Paper IV, showed
comparative concentrations in the inlet to the retention pond in Albertslund
measured by passive sampling, by volume proportional sampling and using a
dynamic stormwater quality model. This indicates that the assumptions and
approximations are reasonable in order to get an estimation of the average
concentrations at the site. However more research is needed in order to evaluate
the accuracy and precision of this type of sampling and to quantify the
uncertainty caused by these assumptions on the uptake of analytes in the flow-
through passive sampler when installed velocity dependant.
39
6 Monitoring of priority pollutants in stormwater runoff using passive samplers and models
The WFD requirements for monitoring are simple and, as discussed in section
2.1 and illustrated in Figure 4 (p. 8), based on a minimum number of grab
samples from surface waters. There are possibilities for using many other
sampling techniques (see Table 1) when designing surface water monitoring
strategies (see also Allan et al. (2006a)). In Figure 13, the role of time and
velocity dependant passive samplers, the passive dosing method and modeling in
WFD monitoring is illustrated.
Figure 13: Possibilities for alternative surface water monitoring strategies.
The most obvious aim where (kinetic) passive samplers can be used as an
alternative sampling method is for surveillance and operational monitoring
(targeting AA-EQS) of surface waters with slow or medium concentration
dynamics. Monthly passive sampler installations would, by integrating the water
concentrations over time, yield more reliable information than monthly grab
samples. Lower detection limits are also often obtained with passive samplers
than with water samples, improving the possibility that detection limits in the
range of the EQSs can be reached. Since the EQSs are defined for total
Surveillance monitoringOperational monitoring
AA-EQS MAC-EQS
Investigative monitoring
Point sources and
stormwater discharges
Grab samples Volume-proportional
samplingDiffusion based
passive sampling
(kinetic, time
dependant)
+ passive dosing
Flow-through
passive sampling
(Velocity dependant)
Lakes, rivers or streams
Monitoring type:
Sampling point:
Sampling method:
Modeling
40
concentrations, a conversion of the dissolved concentrations measured by the
diffusion based passive samplers is needed. The passive dosing method
developed during this PhD can be used for this purpose. This can practically be
done by measuring partitioning in samples taken when the passive samplers are
installed and collected.
The flow-through passive sampler can also be a useful alternative to grab
sampling for surveillance or operational monitoring (targeting AA-EQS) in
flowing surface waters (rivers or streams). Depending on whether the aim of the
sampling is to estimate loads or average concentrations in the surface waters,
time or velocity dependant installations can be chosen. The flow-through passive
sampler can also be dimensioned for low detection limits as required for
comparison with the EQSs.
None of the traditional methods (grab, time or volume proportional sampling) are
suitable for finding the maximum concentrations in surface waters, and passive
samplers are not suitable either, as they have an integrative nature. Sensors or
biological early warning systems are therefore more appropriate if maximum
concentrations should be evaluated (however, sensors are still very expensive).
For investigative monitoring, the flow-through passive sampler can also be a
useful tool. Loads from the sources to the receiving waters can be evaluated
using flow-through passive samplers installed velocity dependently. For a water
body where multiple stormwater discharges contribute to the loads of PPs,
passive sampling is suggested as a cost-efficient alternative to volume
proportional sampling e.g. in order to evaluate the role of each catchment on the
amount of PPs discharged to the water body and to find the best treatment
strategy for the stormwater discharges.
Evaluation of pollutant loads or concentrations in stormwater runoff based on
water quality measurements at a particular site is always based on some kind of
hypothesis or model of the system. The simplest model is to assume that
(sometimes few) measured concentrations represent the general state of the
system. For example, discharge permits can be based on a low number of grab
samples from a site (e.g. 4 per year (Copenhagen Municipality, 2008) or 12 per
year (European Commission, 2000)). If ‘compliance’ or ‘non-compliance’ is
evaluated based on these measurements, representativeness of the samples is
41
assumed even though the permit does not explicitly state this and the supervising
authority may be well aware that this is a coarse simplification. However all
models are simplifications of the system, and it is important to choose the model
according to the aims of the investigation.
Whereas the use of models for design purposes are relatively common, use of
models for permit compliance purposes are rarely seen. An example of this has
been published by Bachhuber (2010), who reported that a modeled annual
average concentration of TSS was accepted by the regulatory authority for
compliance with the discharge permit of an industrial area in Wisconsin.
However, the combination of modeling and sampling for monitoring purposes
has the strong advantage that uncertainty can be lowered as seen in Paper V.
Furthermore, the use of the model can expand the monitoring period beyond the
actually sampled period by evaluating the AA concentration based on longer rain
series and thereby improve the determination of the proper level of treatment for
stormwater discharges.
6.1 Modeling stormwater quality A range of stormwater runoff quality models have been developed over the years
(Elliott and Trowsdale, 2007; Obropta and Kardos, 2007; Tsihrintzis and Hamid,
1997). Some models are simple and some are complex, some models are
deterministic, either conceptual (trying to describe the processes governing
release and transport of pollutants) or regression based (using catchment and/or
rainfall characteristics as parameters), and others are stochastic. The time scale
adopted by the models also differs, where some models targets annual loads,
others are event based, and some are dynamic models with time steps from
minutes to hours.
Examples of regression models can be found in Kayhanian et al. (2007) for
highway runoff constituents and the American national stormwater quality
database (NSQD) and the nationwide urban runoff program (NURP) studies for
mixed stormwater runoff (Maestre and Pitt, 2005; US EPA, 1983). For heavy
metals it has been found that lognormal distributions adequately represent both
the storm to storm variations in pollutant EMCs at an urban site, and the site-to-
site variations in median EMCs (US EPA, 1983). However, since such regression
models require large amounts of data for calibration (Mourad et al., 2005),
42
regression models for the less commonly monitored organic pollutants have not
yet been published.
While regression models are based on data only, conceptual models are
formulated based on knowledge of the system and then calibrated with
measurements in the system. Many conceptual deterministic models have been
described (SWMM, STORM, MOUSE, MUSIC, etc. (Elliott and Trowsdale,
2007; Obropta and Kardos, 2007)). Most of these models are developed for
particulate matter, simulating the buildup and wash-off of particulate matter from
surfaces. These models can be used for pollutants which are sorbed to and
transported by particulate matter and have the same pattern/mechanism of release
from sources in the catchment as the particulate matter. A conceptual
accumulation wash-off model was e.g. calibrated by Lindblom et al. (2011) for
heavy metals. The same model was incorporated in an integrated model for
stormwater systems (Vezzaro et al., 2012). Only few have calibrated conceptual
stormwater quality models for organic pollutants (in Vezzaro et al. (2011) and
Paper V a stormwater quality model was calibrated for fluoranthene), and
application of a stormwater quality model for phthalates and nonylphenols based
on source estimations for input but without calibration showed poor performance
(r2 was 0.01 and 0.03) (Bjorklund et al., 2011).
There are different sources of uncertainty in any model: model structure
uncertainty, parameter uncertainty, uncertainty from the quality and quantity of
calibration data, and input data uncertainty (Butts et al., 2004; Haydon and
Deletic, 2009). Model structure uncertainty is often not evaluated for stormwater
quality models (even though comparisons of models in relation to model
performance have been reported (Dotto et al., 2011)), but model structure
uncertainty have been investigated for hydrological modeling (Butts et al., 2004).
Model parameter uncertainty can be investigated through parameter sensitivity
analysis (Dotto et al., 2011; Vezzaro and Mikkelsen, 2012), input data
uncertainty has e.g. been investigated by Haydon and Deletic (2009) for a
coupled pathogen indicator-hydrologic catchment model, and the influence of the
quantity of data used for calibration have been investigated by e.g. Mourad et al.
(2005). Furthermore, the choice of calibration method can have an effect on
uncertainty estimation (Dotto et al., 2012).
43
Model uncertainty estimation can be done using different methods. Kanso et al.
(2005) showed how a method based on Bayesian theory can be used to estimate
parameter uncertainty in urban runoff quality models. This method consists of
assigning a probability distribution to the model parameters and then updating
the distributions based on the new data collected. Another method used to
estimate model uncertainty is the General Likelihood Uncertainty Estimation
(GLUE) technique (Beven and Binley, 1992). This method is based on the
realization that different parameter sets (called ‘behavioral’) can obtain the same
performance when comparing the model to measurements. The behavioral
parameter sets are chosen based on a likelihood measure which can be chosen by
the modeler. The model prediction bounds are then found from the behavioral
parameter sets. The GLUE method was used for calibration of a dynamic
stormwater quality model by Vezzaro et al. (2012) and in Paper V.
Depending on the use of the models, there are different requirements for the
precision of the models. The more detailed information required from the model,
the more detailed data is necessary for calibration. One approach for gathering
calibration data is to sample ‘mean’ or ‘representative’ rain events (e.g. sampling
for the NURP and NSQD monitoring program includes only ‘representative’
events: “Only samples from rain events greater than 0.1 inches, and close to the
annual mean conditions, were considered valid for the analysis” (Maestre and
Pitt, 2005)). However, it has been shown that calibration of a stormwater quality
regression model on a sub-set of 6 events which included an event with markedly
higher concentrations than the other events significantly improved the accuracy
of the estimate of loads from a catchment compared to a sub-set of events
excluding the high concentration event (Bertrand-Krajewski, 2007). This
highlights the importance of including extreme events in calibration of models. If
the variability of the system should be represented well enough by the model to
evaluate peak concentrations, the data set used for calibration should include
even more detailed information such as pollutographs during the extreme events.
6.2 Use of a dynamic stormwater runoff quality model for evaluating monitoring strategies
In this PhD project, a conceptual dynamic stormwater quality model was used
(Lindblom et al., 2011; Vezzaro, 2011; Vezzaro and Mikkelsen, 2012). The
model consists of a hydrological submodel, where generation and routing of
runoff is simulated, and a water quality submodel where accumulation and
44
release of pollutants is simulated. The runoff model was in Vezzaro et al. (2012)
coupled to a stormwater treatment unit model (Vezzaro et al., 2010). However in
this PhD project only the runoff module was used (Paper V). Calibration of the
model using a general likelihood uncertainty estimation (GLUE) method is
described in Vezzaro et al. (2012), Vezzaro and Mikkelsen (2012) and Paper V.
The aim of the use of the stormwater model was to evaluate the information
gained by different sampling strategies at the catchment in Albertslund.
Therefore simulation of the different sampling methods was implemented in the
model. Volume proportional sampling was simulated by sub-sampling the model
at a given increment in the modeled runoff volume. Passive samplers were
simulated as flow proportional continuous sampling of the modeled
concentration when flow exceeded a given threshold (as described in Paper V).
In reality the sampling rate is related to the water velocity rather than the flow in
the system as discussed in Paper IV. However this was not taken into account in
the model used here which is based on runoff flow and does not include velocity
as a model parameter.
The measurements used for calibration of the model were not chosen to be
‘representative’ or ‘mean’ for the catchment. All sampled events were used.
However, since electricity was not available at the sampling site the samplers had
to be brought home to charge. They were therefore only set up when rain was
expected. This could lead to biases arising from the selection of sampled events
since flashy showers which were not expected in the weather forecasts were not
represented. Very small events were also not represented because the sampled
volume was not enough for analyses. Furthermore, the chance of sampling
extreme events is very low in a sampling campaign consisting of 10 events
sampled over one year. In this respect information gained from passive samplers
has the great advantage that all the events occurring during the installation period
are incorporated in the measurements. The chance of sampling during extreme
events is therefore enhanced. However the robustness of the passive samplers in
extreme events has not been tested, so loss of samplers could occur.
In Figure 14 AA concentrations of Cu and Zn are shown estimated by different
datasets and models. In scenario (a) and (b) a lognormal distribution is assumed
for 6 and 10 events respectively, and the lognormal mean and 50 and 90%
confidence limits are shown (corresponding to the 5, 25, 75 and 95 fractiles).
45
Figure 14: Annual average concentration of Cu (left) and Zn (right) in stormwater runoff in
Albertslund. Mean and 5, 25, 75 and 95% fractiles are estimated using (a) a lognormal
distribution based on 6 EMCs, (b) a lognormal distribution based on 10 EMCs, (f) a dynamic
model based on 6 EMCs and passive sampler measurements. The figure is an extract from
Figure 2 in Paper V.
In scenario (f) the dynamic stormwater quality model was used and calibrated
with 6 EMCs and a passive sampler measurement. The calibration using the
GLUE method results in a number ’behavioral’ parameters set. These parameter
sets are used to evaluate the uncertainty of the model, and in Figure 14 the
median, 5, 25, 75 and 95% fractiles of runs using these parameter sets are shown.
It was seen that using a simple stochastic model based on lognormal distribution
of EMCs lead to high uncertainty on the annual average (AA) concentration in
stormwater estimated from 6 EMCs (scenario a in Figure 14 and in Paper V).
Narrower confidence bounds were obtained by using more events for the
estimation (scenario b in Figure 14). The uncertainty bounds were however
narrowed even further by using the dynamic stormwater quality model for
estimating the AA concentrations. For fluoranthene, where the annual average
was in the range of the current EQSs, the model uncertainty bounds were reduced
to around 50%, which is the performance criteria required for methods of
analysis at the level of the relevant EQS in the WFD (see Paper V and
(European Commission, 2009)).
Bertrand-Krajewski (2007) saw that using different data-sets for calibration of a
pollutant runoff model led to differences in the estimated annual average
0
100
200
300
400
500
600
a b f
µg/
L
Cu Annual Average
0
200
400
600
800
1000
1200
1400
a b f
µg/
L
Zn Annual Average
46
concentrations. This was also to some extent observed at the site investigated in
Paper V. This indicated that the initial data set did not comprise enough events
or appropriate events in order to describe the system fully. The parameters which
are calibrated in the quality model are the pollutant accumulation rate, dry
weather removal rate and rainfall washoff coefficient (see Paper V), and these
parameters are influenced by the hydrological parameters such as runoff
coefficient and initial loss. It was seen that calibration using a combination of
passive sampler measurements and volume proportional samples stabilized the
model by determining an appropriate average level (pollutant accumulation and
dry weather removal rate) and resulted in smaller uncertainty intervals than
calibration based on volume proportional samples only.
The monitoring strategy which resulted in the smallest uncertainty bounds, were
the use of 6 EMCs and one passive sampler measurement to calibrate the
dynamic stormwater quality model (as seen in scenario (f) in Figure 14 and in
Paper V). This strategy minimizes the measurement costs since passive sampling
is less costly than volume proportional sampling.
47
7 Conclusions This PhD thesis provides an overview of important issues for monitoring of PPs
in stormwater. The most important PPs identified in stormwater were the heavy
metals Cu and Zn, PAHs, DEHP and the pesticides used locally (here
glyphosate). These PPs were found in all of the 6 stormwater screening samples
from different locations in the Copenhagen area. These PPs also represent the
main stormwater relevant pollutant groups and sources found in literature.
Partitioning of pollutants between particulate matter and freely dissolved forms
in stormwater is important for the toxicity of PPs, for treatment of PPs, but also
for sampling of stormwater because e.g. most passive samplers only sample the
dissolved and labile forms. The sorption capacity of stormwater can be
characterized using a novel analytical method which was developed during this
PhD project for measuring partitioning of hydrophobic organic compounds in
aqueous samples. This method proved to be simple to use and provide precise
and robust partition measurements. The method revealed a logKTSS of 4.59 for
fluoranthene in stormwater from a catchment in Denmark and another in France.
Calculations based on the partitioning measurements of fluoranthene and
inherent properties of the other PAHs showed the different partitioning of the
PAHs in stormwater, with naphthalene being mainly freely dissolved in
stormwater, fluoranthene being mainly sorbed to particulate matter during the
first part of the runoff events but up to 50 % freely dissolved during the last part
of events, and benzo(a)pyrene being mainly sorbed to particulate matter in
stormwater.
Most of the passive samplers described in literature cannot be used reasonably in
stormwater systems since time-weighted averages over weeks to months do not
yield useful information in systems with long periods of no-flow. However, a
flow-through passive sampler, SorbiCell, can be used because it can be installed
in a manner where sampling is dependent on the velocity of the water in the
system thus facilitating sampling primarily during runoff events. Tests of this
sampler in a stormwater drainage tunnel revealed concentrations of heavy metals
comparable to total concentrations measured by volume proportional sampling
and to flow-weighted concentrations obtained using a dynamic stormwater
quality model. The advantage of using this passive sampler is the integrative
nature of the sampling, that sampling is not limited by the availability of
personnel during odd hours, and that transport of heavy and large equipment and
48
large sample volumes is avoided. The disadvantages are that the samplers are not
validated yet and that the sampling uncertainties and biases have not been fully
quantified. Furthermore there are technical challenges such as avoiding blockage
by leaves and debris and ensuring a high water velocity at the sampler
installation.
Many stormwater quality models exists, however there is no consensus on how
much information is needed to evaluate mean concentrations at site level. The
advantage of using models for monitoring purposes is that information about the
system beyond the time interval of sampling can be obtained based on
knowledge of processes and observed patterns. Furthermore, statistics on the
annual average concentrations can be obtained from modeling of rain series.
Model prediction bounds for annual average concentrations obtained by a
dynamic stormwater quality model were narrower than using simple stochastic
methods. The use of passive sampler measurements in combination with volume
proportional measurements for calibration reduced the model prediction bounds
on annual average concentrations more than simply increasing the number of
volume proportional samples.
The work done during this PhD project illustrates a framework for monitoring
where passive samplers and models are used. It is an important step towards
general acceptance and use of passive samplers and models in monitoring
programs for stormwater runoff. This can provide practical and economical
advantages compared to traditional methods and can enable larger scale
monitoring of PPs in stormwater to be feasible.
49
8 Future research Due to the very high ambitions outlined by the EU WFD for the water quality of
surface waters, including phasing out of emissions of the traffic related PAHs,
urban stormwater management is crucial. The limited knowledge on emissions of
some of the priority pollutants to surface waters necessitates monitoring
campaigns. As many of the EQSs defined are lower than the detection limits of
the standard analytical methods used by commercial laboratories, there is a great
need for development of analytical methods to detect pollutants in pg/L
concentrations.
In this study the flow-through passive sampler, SorbiCell, was tested in
stormwater runoff. Even though the test showed promising results for velocity
dependant sampling of stormwater runoff, work is still needed in order to
understand and describe the sampling method. In order to gain more information
about the sampler, lab-scale experiments can be designed to investigate:
- the relationship between the water velocity and the flow through the
sampler,
- possible decrease in response over time due to clogging for long-time
installations,
- the uptake of analytes by the sampler during stagnant flow-conditions,
- the pore size of the spheriglass filter.
There is also a need for standardized installation methods which avoids problems
with blocking of the sampler by leaves etc. and ensures controlled flow
conditions. Furthermore, even though many passive samplers have detection
limits in the pg/L range (depending on the actual installation), the tested flow-
through passive sampler had higher detection limits in this work. Optimization of
the passive sampler for lower detection limits is important in order to be able to
use it for monitoring of organic pollutants in stormwater runoff. The use of flow-
through passive samplers for pesticide monitoring also has great potential, and
development of a resin for binding of hydrophilic pesticides such as glyphosate
would be useful.
More measurements of the partitioning in stormwater samples would also be very
interesting in order to see whether the partition ratios found in this study are in
50
fact as globally applicable as indicated by the results obtained so far. With the
passive dosing method developed here, these measurements are straight forward.
More generally there is a need to accredit the passive samplers and define their
role in monitoring. Inter-laboratory analysis tests and installations of different
types of passive samplers at the same site are currently being done and enable
comparisons between sampler performances. However there is a need for a
standard method to evaluate the biases and uncertainties of passive samplers at
relevant concentration levels and field conditions similar to the use of reference
materials in standard quality control of laboratory analyses.
Further development of stormwater quality models is also needed. Especially the
ability of the model to predict extreme events and dynamics throughout the
events needs to be investigated further in order to improve the ability of the
models to relate to MAC-EQSs. Criteria for uncertainty and calibration of
models when used for compliance purposes also need to be defined. Future
research should include larger monitoring programs using passive samplers over
long periods, volume proportional sampling over short periods, and stormwater
quality models to evaluate the statistical features of AA concentrations and
maximum concentrations in stormwater discharges.
51
9 References Ackerman, D., Stein, E. D. and Ritter, K. J. (2011). Evaluating performance of stormwater
sampling approaches using a dynamic watershed model. Environmental Monitoring and
Assessment, 180 (1-4), 283-302.
Allan, I. J., Booij, K., Paschke, A., Vrana, B., Mills, G. A. and Greenwood, R. (2010). Short-
term exposure testing of six different passive samplers for the monitoring of hydrophobic
contaminants in water. Journal of Environmental Monitoring, 12 (3), 696-703.
Allan, I. J., Booij, K., Paschke, A., Vrana, B., Mills, G. A. and Greenwood, R. (2009). Field
Performance of Seven Passive Sampling Devices for Monitoring of Hydrophobic
Substances. Environmental Science & Technology, 43 (14), 5383-5390.
Allan, I. J., Vrana, B., Greenwood, R., Mills, G. A., Knutsson, J., Holmberg, A., Guigues, N.,
Fouillac, A. M. and Laschi, S. (2006a). Strategic monitoring for the European Water
Framework Directive. Trends in Analytical Chemistry, 25 (7), 704-715.
Allan, I. J., Vrana, B., Greenwood, R., Mills, G. A., Roig, B. and Gonzalez, C. (2006b). A
"toolbox" for biological and chemical monitoring requirements for the European Union's
Water Framework Directive. Talanta, 69 (2), 302-322.
Alvarez, D. A., Petty, J. D., Huckins, J. N., Jones-Lepp, T. L., Getting, D. T., Goddard, J. P. and
Manahan, S. E. (2004). Development of a passive, in situ, integrative sampler for
hydrophilic organic contaminants in aquatic environments. Environmental Toxicology and
Chemistry, 23 (7), 1640-1648.
Arthur, C. L. and Pawliszyn, J. (1990). Solid-phase microextraction with thermal-desorption
using fused-silica optical fibers. Analytical Chemistry, 62 (19), 2145-2148.
Bachhuber, J. A. (2010). Managing stormwater pollutants from a heavy manufacturing site:
Monitoring, modeling, and design. Watershed Management Conference 2010: Innovations
in Watershed Management under Land Use and Climate Change - Proceedings of the
2010 Watershed Management Conference, 394, 670-682.
Baltussen, E., Sandra, P., David, F. and Cramers, C. (1999). Stir bar sorptive extraction (SBSE),
a novel extraction technique for aqueous samples: Theory and principles. Journal of
Microcolumn Separations, 11 (10), 737-747.
Battaglin, W. A. and Hay, L. E. (1996). Effects of sampling strategies on estimates of annual
mean herbicide concentrations in midwestern rivers. Environmental Science and
Technology, 30 (3), 889-896.
Baun, A., Eriksson, E., Ledin, A. and Mikkelsen, P. S. (2008). A methodology for ranking and
hazard identification of xenobiotic organic compounds in urban stormwater. Science of the
Total Environment, 370, 29-38.
Bertrand-Krajewski, J. L. (2007). Stormwater pollutant loads modelling: Epistemological
aspects and case studies on the influence of field data sets on calibration and verification.
Water Science and Technology, 55 (4), 1-17.
52
Bertrand-Krajewski, J. L., Becouze, C., Dembélé, A., Coquery, M., Cren-Olivé, C., Barillon, B.,
Dauthuille, P., Chapgier, J., Grenier-Loustalot, M.-F. and Marin, P. (2008). Priority
pollutants in stormwater: The 'ESPRIT' project. Proceedings of the 11th International
conference on Urban Drainage.
Beven, K. and Binley, A. (1992). The future of distributed models - Model calibration and
uncertainty prediction. Hydrological processes, 6 (3), 279-298.
Bjorklund, K., Malmqvist, P.-A. and Stromvall, A.-M. (2011). Simulating organic pollutant
flows in urban stormwater: development and evaluation of a model for nonylphenols and
phthalates. Water Science and Technology, 63 (3), 508-515.
Blom, L. B., Morrison, G. M., Kingston, J., Mills, G. A., Greenwood, R., Pettersson, T. J. R.
and Rauch, S. (2002). Performance of an in situ passive sampling system for metals in
stormwater. Journal of Environmental Monitoring, 4 (2), 258-262.
Booij, K., Smedes, F. and van Weerlee, E. M. (2002). Spiking of performance reference
compounds in low density polyethylene and silicone passive water samplers.
Chemosphere, 46 (8), 1157-1161.
Brannon, J. M., Pennington, J. C., Davis, W. M. and Hayes, C. (1995). Fluoranthene KDOC in
sediment pore waters. Chemosphere, 30, 419-428.
Burkhardt, M., Zuleeg, S., Vonbank, R., Schmid, P., Hean, S., Lamani, X., Bester, K. and
Boller, M. (2011). Leaching of additives from construction materials to urban storm water
runoff. Water Science and Technology, 63 (9), 1974-1982.
Butts, M. B., Payne, J. T., Kristensen, M. and Madsen, H. (2004). An evaluation of the impact
of model structure on hydrological modelling uncertainty for streamflow simulation.
Journal of Hydrology, 298 (1-4), 242-266.
Clark, S. E., Steele, K. A., Steele, K. A., Spicher, J., Siu, C. S., Lalor, M. M., Pitt, R. and Kirby,
J. T. (2008). Roofing materials' contributions to storm-water runoff pollution. Journal of
Irrigation and Drainage Engineering, 134 (5), 638-645.
Copenhagen Municipality. Midlertidig udledningstilladelse til Hovedgrøften på Amager Fælled,
AFU03 (Temporary discharge permit for Hovedgrøften at Amager Fælled, AFU03).
Sagsnr. 2007-93221, Dokumentnr. 2007-387212. 2008.
Danish Environmental Protection Agency (2000) NOVA-2003, Programbeskrivelse for det
nationale program for overvågning af vandmiljøet i Danmark, 1998-2003. Redegørelse 1.
Ministry of the Environment,
Danish Ministry of the Environment (2010). Bekendtgørelse om miljøkvalitetskrav for
vandområder og krav til udledning af forurenende stoffer til vandløb, søer eller havet
(Executive order on environmental quality criteria for water bodies and requirements for
discharges of pollutants to water streams, lakes and the sea). Executive order no. 1022
from 25/08/2010.
Davison and Zhang (1994). In situ speciation measurements of trace components in natural
waters using thin-film gels. Nature, 367 (6463), 546-548.
53
de Jonge, H. and Rothenberg, G. (2005). New Device and Method far Flux-Proportional
Sampling of Mobile Salutes in Soil and Groundwater. Environmental Science &
Technology, 39 (1).
de Maagd, P. G.-J., Sinnige, T. L., Schrap, S. M., Opperhuizen, A. and Sijm, D. T. H. M.
(1998). Sorption coefficients of polycyclic aromatic hydrocarbons for two lake sediments:
Influence of the bactericide sodium azide. Environmental Toxicology and Chemistry, 17
(10), 1899-1907.
Dotto, C. B. S., Kleidorfer, M., Deletic, A., Rauch, W., McCarthy, D. T. and Fletcher, T. D.
(2011). Performance and sensitivity analysis of stormwater models using a Bayesian
approach and long-term high resolution data. Environmental Modelling and Software, 26
(10), 1225-1239.
Dotto, C. B. S., Mannina, G., Kleidorfer, M., Vezzaro, L., Henrichs, M., McCarthy, D. T.,
Freni, G., Rauch, W. and Deletic, A. (2012). Comparison of different uncertainty
techniques in urban stormwater quantity and quality modelling. Water Research, 46 (8),
2545-2558.
Elliott, A. H. and Trowsdale, S. A. (2007). A review of models for low impact urban stormwater
drainage. Environmental Modelling & Software, 22 (3), 394-405.
Eriksson, E., Baun, A., Mikkelsen, P. S. and Ledin, A. (2005). Chemical hazard identification
and assessment tool for evaluation of stormwater priority pollutants. Water Science and
Technology, 51 (2), 47-55.
Eriksson, E., Baun, A., Mikkelsen, P. S. and Ledin, A. (2007a). Risk assessment of xenobiotics
in stormwater discharged to Harrestrup Å, Denmark. Desalination, 215, 187-197.
Eriksson, E., Baun, A., Scholes, L., Ledin, A., Ahlman, S., Revitt, M., Noutsopoulos, C. and
Mikkelsen, P. S. (2007b). Selected stormwater priority pollutants - a European
perspective. Science of the Total Environment, 383, 41-51.
European Commission (2000). Directive 2000/60/EC of the European Parliament and of the
Council of 23 October 2000 establishing a framework for Community action in the field of
water policy.
European Commission (2011). Proposal for a directive of the european parliament and of the
council amending Directives 2000/60/EC and 2008/105/EC as regards priority substances
in the field of water policy.
European Commission (2009). Commission directive 2009/90/EC of 31 July 2009, laying
down, pursuant to Directive 2000/60/EC of the European Parliament and of the Council,
technical specifications for chemical analysis and monitoring of water status.
European Commission (2008). Directive 2008/105/EC of the European Parliament and of the
Council of 16 December 2008 on environmental quality standards in the field of water
policy.
Fernandes, M. B., Sicre, M. A., Boireau, A. and Tronczynski, J. (1997). Polyaromatic
Hydrocarbon (PAH) Distributions in the Seine River and its Estuary. Marine Pollution
Bulletin, 34 (11), 857-867.
54
Fletcher, T. D. and Deletic, A. (2007). Statistical evaluation and optimisation of stormwater
quality monitoring programmes. Water Science and Technology, 56 (12), 1-9.
Gasperi, J., Garnaud, S., Rocher, V. and Moilleron, R. (2009). Priority pollutants in surface
waters and settleable particles within a densely urbanised area: Case study of Paris
(France). Science of the Total Environment, 407 (8), 2900-2908.
Gauthier, T. D., Seitz, W. R. and Grant, C. L. (1987). Effects of structural and compositional
variations of dissolved humic materials on pyrene Koc values. Environmental Science and
Technology, 21 (3), 243-248.
Gauthier, T. D., Shane, E. C., Guerin, W. F., Seitz, W. R. and Grant, C. L. (1986). Fluorescence
quenching method for determining equilibrium constants for polycyclic aromatic
hydrocarbons binding to dissolved humic materials. Environmental Science &
Technology, 20 (11), 1162-1166.
Gettel, M., Gulliver, J. S., Kayhanian, M., DeGroot, G., Brand, J., Mohseni, O. and Erickson, A.
J. (2011). Improving suspended sediment measurements by automatic samplers. Journal
of Environmental Monitoring, 13 (10), 2703-2709.
Göbel, P., Dierkes, C. and Coldewey, W. G. (2007). Storm water runoff concentration matrix
for urban areas. Journal of Contaminant Hydrology, 91 (1-2), 26-42.
Greenwood,R., Mills,G., and Vrana,B. (2007) Comprehensive Analytical Chemistry, Passive
Sampling Techniques in Environmental Monitoring., Elsevier B.V., Amsterdam, The
Netherlands.
Haftka, J. J. H., Govers, H. A. J. and Parsons, J. R. (2010). Influence of temperature and origin
of dissolved organic matter on the partitioning behavior of polycyclic aromatic
hydrocarbons. Environmental Science and Pollution Research International, 17 (5), 1070-
1079.
Hawthorne, S. B., Grabanski, C. B. and Miller, D. J. (2006). Measured partitioning coefficients
for parent and alkyl polycyclic aromatic hydrocarbons in 114 historically contaminated
sediments: Part 1. KOC values. Environmental Toxicology and Chemistry, 25 (11), 2901-
2911.
Haydon, S. and Deletic, A. (2009). Model output uncertainty of a coupled pathogen indicator-
hydrologic catchment model due to input data uncertainty. Environmental Modelling and
Software, 24 (3), 322-328.
Hazelton, C. (1998). Variations between continuous and spot-sampling techniques in
monitoring a change in river-water quality. Journal of the chartered institution of water
and environmental management, 12 (2), 124-129.
He, Y., Yediler, A., Sun, T. and Kettrup, A. (1995). Adsorption of fluoranthene on soil and lava:
effects of the organic carbon contents of adsorbents and temperature. Chemosphere, 30
(1), 141-150.
Horowitz, A. J. (2008). Determining annual suspended sediment and sediment-associated trace
element and nutrient fluxes. Science of the Total Environment, 400 (1-3), 315-343.
55
Huckins, J. N., Manuweera, G. K., Petty, J. D., Mackay, D. and Lebo, J. A. (1993). Lipid-
Containing Semipermeable-Membrane Devices for Monitoring Organic Contaminants in
Water. Environmental Science & Technology, 27 (12), 2489-2496.
Huckins,J.N., Petty,J.D., and Booij,K. (2006) Monitors of organic Chemicals in the
Environment., Springer, New York.
Huckins, J. N., Tubergen, M. W. and Manuweera, G. K. (1990). Semipermeable membrane
devices containing model lipid: A new approach to monitoring the bioavaiiability of
lipophilic contaminants and estimating their bioconcentration potential. Chemosphere, 20
(5), 533-552.
Ingvertsen, S. T., Jensen, M. B. and Magid, J. (2011). A Minimum Data Set of Water Quality
Parameters to Assess and Compare Treatment Efficiency of Stormwater Facilities. Journal
of Environmental Quality, 40 (5), 1488-1502.
Jonker, M. T. O., Hawthorne, S. B. and Koelmans, A. A. (2005). Extremely slowly desorbing
polycyclic aromatic hydrocarbons from soot and soot-like materials: Evidence by
supercritical fluid extraction. Environmental Science & Technology, 39 (20), 7889-7895.
Jonker, M. T. O. and Koelmans, A. A. (2002). Sorption of polycyclic aromatic hydrocarbons
and polychlorinated biphenyls to soot and soot-like materials in the aqueous environment:
Mechanistic considerations. Environmental Science and Technology, 36 (17), 3725-3734.
Kanso, A., Chebbo, G. and Tassin, B. (2005). Stormwater quality modelling in combined
sewers: calibration and uncertainty analysis. Water Science and Technology, 52 (3), 63-71.
Kayhanian, M., Suverkropp, C., Ruby, A. and Tsay, K. (2007). Characterization and prediction
of highway runoff constituent event mean concentration. Journal of Environmental
Management, 85 (2), 279-295.
Kingston, J. K., Greenwood, R., Mills, G. A., Morrison, G. M. and Bj+Ârklund Persson, L.
(2000). Development of a novel passive sampling system for the time-averaged
measurement of a range of organic pollutants in aquatic environments. Journal of
Environmental Monitoring, 2 (5), 487-495.
Kjølholt, J., Skaarup, S., and Ringgaard, K. W. (2007) Basisviden om EU-regulerede stoffer i
vandmiljøet, regulering, anvendelser, forureningskilder og forekomst (Basic knowledge on
EU-regulated substances in the aquatic environment, regulation, uses, sources and
occurence). Environmental project 1181. Danish Environmental Protection Agency,
Available at: http://www2.mst.dk/Udgiv/publikationer/2007/978-87-7052-563-3/pdf/978-
87-7052-564-0.pdf (Accessed 14/7 11 A.D.).
Komarova, T., Bartkow, M. E., Muller, J. F., Carter, S. and Vanderzalm, J. (2006). Field
evaluation of passive samplers: Monitoring polycyclic aromatic hydrocarbons (PAHs) in
stormwater. Polycyclic Aromatic Compounds, 26 (3), 221-236.
Kronvang, B., Hoffmann, C., Kjærsgaard, C., de Jonge, H. and Audet, J. (2010). Hvorfor
overvåge reetablerede vådområder? (Why monitor reestablished wetlands?). Vand og Jord
(Water and Soil), 17 (2), 67-72.
56
Kumata, H., Masuda, K., Yamada, J. and Takada, H. (2000). Water-particle distribution of
hydrophobic micro pollutants in storm water runoff. Polycyclic Aromatic Compounds, 20
(1-4), 39-54.
Landrum, P. F., Nihart, S. R., Eadie, B. J. and Gardner, W. S. (1984). Reverse-phase separation
method for determining pollutant binding to Aldrich humic acid and dissolved organic
carbon of natural waters. Environmental Science & Technology, 18 (3), 187-192.
Laor, Y. and Rebhun, M. (1997). Complexation-flocculation: A new method to determine
binding coefficients of organic contaminants to dissolved humic substances.
Environmental Science and Technology, 31 (12), 3558-3564.
Ledin, A., Auffarth, K., Boe-Hansen, R., Eriksson, E., Albrechtsen, H.-J., Baun, A., and
Mikkelsen, P. S. (2004) Brug af regnvand opsamlet fra tage og befæstede arealer (Use of
rain water collected from roofs and paved areas). Økolgisk byfirnyelse og
spildevandsrensning (Ecological urban regeneration and wastewater treatment) 48. Danish
Ministry of the Environment, Available at:
http://www2.mst.dk/Udgiv/publikationer/2004/87-7614-153-5/pdf/87-7614-154-3.pdf
(Accessed 31/5 2012).
Lee, H., Swamikannu, X., Radulescu, D., Kim, S. J. and Stenstrom, M. K. (2007). Design of
stormwater monitoring programs. Water Research, 41 (18), 4186-4196.
Lee, S., Gan, J., Liu, W. P. and Anderson, M. A. (2003). Evaluation of K-d underestimation
using solid phase microextraction. Environmental Science & Technology, 37 (24), 5597-
5602.
Leecaster, M. K., Schiff, K. and Tiefenthaler, L. L. (2002). Assessment of efficient sampling
designs for urban stormwater monitoring. Water Research, 36 (6), 1556-1564.
Lindblom, E., Ahlman, S. and Mikkelsen, P. S. (2011). Uncertainty-based calibration and
prediction with a stormwater surface accumulation-washoff model based on coverage of
sampled Zn, Cu, Pb and Cd field data. Water Research, 45 (13), 3823-3835.
Luers, F. and tenHulscher, T. E. M. (1996). Temperature effect on the partitioning of polycyclic
aromatic hydrocarbons between natural organic carbon and water. Chemosphere, 33 (4),
643-657.
Lützhøft, H.-C. H., Donner, E., Wickman, T., Eriksson, E., Banovec, P., Mikkelsen, P. S. and
Ledin, A. (2012). A source classification framework supporting pollutant source mapping,
pollutant release prediction, transport and load forecasting, and source control planning for
urban environments. Environmental science and pollution research international., 19 (4),
1119-1130.
Ma, J. S., Kang, J. H., Kayhanian, M. and Stenstrom, M. K. (2009). Sampling issues in Urban
runoff monitoring programs: Composite versus grab. Journal of Environmental
Engineering, 135 (3), 118-127.
Maestre, A. and Pitt, R. (2005) The National Stormwater Quality Database, Version 1.1, A
Compilation and Analysis of NPDES Stormwater Monitoring Information. US EPA,
Office of Water, Washington D.C, USA. Available at:
http://rpitt.eng.ua.edu/Publications/Stormwater%20Characteristics/NSQD%20EPA.pdf
(Accessed 10/11 2011).
57
Mazzella, N., Dubernet, J. F. and Delmas, F. (2007). Determination of kinetic and equilibrium
regimes in the operation of polar organic chemical integrative samplers - Application to
the passive sampling of the polar herbicides in aquatic environments. Journal of
Chromatography A, 1154 (1-2), 42-51.
McGroddy, S. E. and Farrington, J. W. (1995). Sediment porewater partitioning of polycyclic
aromatic hydrocarbons in three cores from Boston Harbor, Massachusetts. Environmental
Science and Technology, 29 (6), 1542-1550.
Mills, G. A., Greenwood, R., Vrana, B., Allan, I. J. and Ocelka, T. (2011). Measurement of
environmental pollutants using passive sampling devices - a commentary on the current
state of the art. Journal of Environmental Monitoring, 13 (11), 2979-2982.
Mourad, M., Bertrand-Krajewski, J. L. and Chebbo, G. (2005). Sensitivity of stormwater quality
models to experimental data used for calibration. Houille Blanche-Revue Internationale de
l Eau, (6), 68-74.
Müller, J. F., Manomanii, K., Mortimer, M. R. and McLachlan, M. S. (2001). Partitioning of
polycyclic aromatic hydrocarbons in the polyethylene/water system. Fresenius' journal of
analytical chemistry., 371 (6), 816-822.
Namiesnik, J., Zabiegala, B., Kot-Wasik, A., Partyka, M. and Wasik, A. (2005). Passive
sampling and/or extraction techniques in environmental analysis: a review. Analytical and
Bioanalytical Chemistry, 381 (2), 279-301.
Neale, P. A., Antony, A., Gernjak, W., Leslie, G. and Escher, B. I. (2011). Natural versus
wastewater derived dissolved organic carbon: Implications for the environmental fate of
organic micropollutants. Water Research, 45 (14), 4227-4237.
Obropta, C. C. and Kardos, J. S. (2007). Review of urban stormwater quality models:
Deterministic, stochastic, and hybrid approaches. Journal of the American Water
Resources Association, 43 (6), 1508-1523.
Ort, C., Gujer, W., Ort, C. and Gujer, W. (2006). Sampling for representative micropollutant
loads in sewer systems. Water Science and Technology, 54 (6-7), 169-176.
Persson, L. B., Morrison, G. M., Friemann, J. U., Kingston, J., Mills, G. and Greenwood, R.
(2001). Diffusional behaviour of metals in a passive sampling system for monitoring
aquatic pollution. Journal of Environmental Monitoring, 3 (6), 639.
Persson, N. J., Bucheli, T. D., Gustafsson, Ö., Broman, D., Næs, K., Ishaq, R. and Zebühr, Y.
(2005). Testing common sediment porewater distribution models for their ability to predict
dissolved concentrations of POPs in The Grenlandsfjords, Norway. Chemosphere, 59 (10),
1475-1485.
Pettersson, M., de Keyser, W., Birch, H., Lützhøft, H.-C. H., Gevaert, V., Cerk, M., Benedetti,
l., and Vanrolleghem, P. (2010) Strategies for monitoring of priority pollutant emission
barriers. ScorePP Deliverable D7.5 a Specific Targeted Reserach Project funded by the
European Comission under the Sixth Framework Programme, Available at:
www.scorepp.eu
58
Puchalski, M. M., Morra, M. J. and VON, W. R. (1992). Fluorescence quenching of synthetic
organic compounds by humic materials. Environmental Science and Technology, 26 (9),
1787-1792.
Raber, B., Kogel-Knabner, I., Stein, C. and Klem, D. (1998). Partitioning of polycyclic aromatic
hydrocarbons to dissolved organic matter from different soils. Chemosphere, 36 (1), 79-
97.
Roseen, R. M., Ballestero, T. P., Fowler, G. D., Guo, Q. and Houle, J. (2011). Sediment
monitoring bias by automatic sampler in comparison with large volume sampling for
parking lot runoff. Journal of Irrigation and Drainage Engineering, 137 (4), 251-257.
Rozemeijer, J., van der Velde, Y., de Jonge, H., van Geer, F., Broers, H. P. and Bierkens, M.
(2010). Application and Evaluation of a New Passive Sampler for Measuring Average
Solute Concentrations in a Catchment Scale Water Quality Monitoring Study.
Environmental Science & Technology, 44 (4), 1353-1359.
Schoknecht, U., Gruycheva, J., Mathies, H., Bergmann, H. and Burkhardt, M. (2009). Leaching
of Biocides Used in Facade Coatings under Laboratory Test Conditions. Environmental
Science & Technology, 43 (24), 9321-9328.
Sebastian, C., Moilleron, R., Ruban, V., Barraud, S., Lorgeoux, C., Saad, M., Cren, C., Wiest,
L., Percot, S., Demare, D., Gromaire, M.-C., Gasperi, J., Millet, M., Maro, D. and Chebbo,
G. (2011). INOGEV project – an original French approach in micropollutant
characterization assessment in urban wet weather effluents and atmospheric deposits.
Proceedings of the 12th International Conference on Urban Drainage.
Seethapathy, S., Gorecki, T. and Li, X. (2008). Passive sampling in environmental analysis.
Journal of Chromatography A, 1184 (1-2), 234-253.
Skarzynska, K., Polkowska, Z., Namiesnik, J. and Przyjazny, A. (2007). Application of different
sampling procedures in studies of composition of various types of runoff waters - A
review. Critical Reviews in Analytical Chemistry, 37 (2), 91-105.
Sörme, L. and Lagerkvist, R. (2002). Sources of heavy metals in urban wastewater in
Stockholm. Science of the Total Environment, 298 (1-3), 131-145.
Stuer-Lauridsen, F. (2005). Review of passive accumulation devices for monitoring organic
micropollutants in the aquatic environment. Environmental Pollution, 136 (3), 503-524.
Ter Laak, T. L., Hermens, J. L. M., Durjava, M., Struijs, J. and Durjava, M. (2005). Solid phase
dosing and sampling technique to determine partition coefficients of hydrophobic
chemicals in complex matrixes. Environmental Science and Technology, 39 (10), 3736-
3742.
Tran, A. T. K., Hyne, R. V. and Doble, P. (2007). Calibration of a passive sampling device for
time-integrated sampling of hydrophilic herbicides in aquatic environments.
Environmental Toxicology and Chemistry, 26 (3), 435-443.
Tremblay, L., Kohl, S. D., Rice, J. A. and Gagne, J. P. (2005). Effects of temperature, salinity,
and dissolved humic substances on the sorption of polycyclic aromatic hydrocarbons to
estuarine particles. Marine Chemistry, 96 (1-2), 21-34.
59
Tsihrintzis, Vassilios A. and Hamid, Rizwan. Modeling and management of urban stormwater
runoff quality: a review. Water Resources Management 11[2], 137-164. 1997. Kluwer
Academic Publishers.
US EPA (1983) Final Report of the Nationwide Urban Runoff Program. Office of Water, U.S.
Environmental Protection Agency, Washington D.C. 20460.
US EPA (1992) NPDES Storm Water Sampling Guidance Document. EPA 833-B-92-001.
Office of wastewater enforcement and compliance, Washington, DC. Available at:
http://www.epa.gov/npdes/pubs/owm0093.pdf (Accessed 1/11 2012).
US EPA. Elements of a State Water Monitoring and Assessment Program. EPA 841-B-03-003 .
2003.
van Leeuwen, H. P., Town, R. M., Buffle, J., Cleven, R. F. M. J., Davison, W., Puy, J., Van
Riemsdijk, W. H. and Sigg, L. (2005). Dynamic speciation analysis and bioavailability of
metals in aquatic systems. Environmental Science & Technology, 39 (22), 8545-8556.
Vezzaro, L. (2011) Source-Flux-Fate Modelling of Priority Pollutants in Stormwater Systems.
PhD-thesis, Department of Environmental Engineering, Technical University of
Denmark.
Vezzaro, L., Eriksson, E., Ledin, A. and Mikkelsen, P. S. (2010). Dynamic stormwater
treatment unit model for micropollutants (STUMP) based on substance inherent
properties. Water Science and Technology, 62 (3), 622-629.
Vezzaro, L., Ledin, A. and Mikkelsen, P. S. (2012). Integrated modelling of Priority Pollutants
in stormwater systems. Physics and Chemistry of the Earth, 42-44, 42-51.
Vezzaro, L. and Mikkelsen, P. S. (2012). Application of global sensitivity analysis and
uncertainty quantification in dynamic modelling of micropollutants in stormwater runoff.
Environmental Modelling and Software, 27-28, 40-51.
Vezzaro, L., Sharma, A. K., Ledin, A. and Mikkelsen, P. S. (2011). Evaluating stormwater
micropollutant control strategies by the application of an integrated model. Proceedings of
the 12th International Conference on Urban Drainage.
Vrana, B., Mills, G. A., Allan, I. J., Dominiak, E., Svensson, K., Knutsson, J., Morrison, G. and
Greenwood, R. (2005). Passive sampling techniques for monitoring pollutants in water.
Trends in Analytical Chemistry, 24 (10), 845-868.
Vrana, B., Mills, G. A., Kotterman, M., Leonards, P., Booij, K. and Greenwood, R. (2007).
Modelling and field application of the Chemcatcher passive sampler calibration data for
the monitoring of hydrophobic organic pollutants in water. Environmental Pollution, 145
(3), 895-904.
Wicke, D., Boeckelmann, U. and Reemtsma, T. (2007). Experimental and modeling approach to
study sorption of dissolved hydrophobic organic contaminants to microbial biofilms.
Water Research, 41 (10), 2202-2210.
60
Witt, G., Bartsch, C., Liehr, G. A., Thiele, R. and McLachlan, M. S. (2010). Using solid-phase
microextraction to evaluate the role of different carbon matrices in the distribution of
PAHs in sediment-porewater systems of the Baltic Sea. Journal of Soils and Sediments, 10
(7), 1388-1400.
Worms, I., Simon, D. F., Hassler, C. S. and Wilkinson, K. J. (2006). Bioavailability of trace
metals to aquatic microorganisms: importance of chemical, biological and physical
processes on biouptake. Biochimie, 88 (11), 1721-1731.
Zgheib, S., Moilleron, R., Saad, M. and Chebbo, G. (2011). Partition of pollution between
dissolved and particulate phases: What about emerging substances in urban stormwater
catchments? Water Research, 45 (2), 913-925.
Zhang, H. and Davison, W. (1995). Performance characteristics of diffusion gradients in thin
films for the in situ measurement of.. Analytical Chemistry, 67 (19).
Zhang, Z., Hibberd, A. and Zhou, J. L. (2008). Analysis of emerging contaminants in sewage
effluent and river water: Comparison between spot and passive sampling. Analytica
Chimica Acta, 607 (1), 37-44.
Zhou, J. L., Fileman, T. W., Evans, S., Donkin, P., Readman, J. W., Mantoura, R. F. C. and
Rowland, S. (1999). The partition of fluoranthene and pyrene between suspended particles
and dissolved phase in the Humber Estuary: a study of the controlling factors. Science of
the Total Environment, 244, 305-321.
61
10 Papers
I. Birch, H., Mikkelsen, P.S., Jensen, J.K., Lützhøft, H.-C.H. (2011).
Micropollutants in stormwater run-off and combined sewer overflow in
the Copenhagen area, Denmark, Water Science and Technology, 64 (2),
485-493. DOI: 10.2166/wst.2011.687.
II. Birch, H., Gouliarmou, V., Lützhøft, H.-C.H., Mikkelsen, P.S. and
Mayer, P. (2010). Passive Dosing to determine the speciation of
hydrophobic organic chemicals in aqueous samples, Analytical Chemistry,
82 (3), 1142-1146. DOI: 10.1021/ac902378w.
III. Birch, H., Mayer, P., Lützhøft, H-C.H. and Mikkelsen, P.S. (in revision)
Partitioning of fluoranthene between free and bound forms in stormwater
runoff and urban waste waters using passive dosing, Water Research.
IV. Birch, H., Sharma, A.K., Vezzaro, L., Lützhøft, H.-C.H. and Mikkelsen,
P.S. (manuscript). Velocity dependant sampling of micropollutants in
stormwater using a flow-through passive sampler. Environmental Science
and Technology. 2012
V. Birch, H., Vezzaro, L. and Mikkelsen, P.S. (accepted) Model based
monitoring of stormwater runoff quality. In proceedings of the 9th
international conference on Urban Drainage Modelling, Belgrade, 2012.
The papers are not included in the www-version, but can be obtained from the
Library at DTU Environment.
Department of Environmental Engineering
Technical University of Denmark
Miljoevej, Building 113
DK-2800 Kgs. Lyngby, Denmark
Technical University of Denmark
DTU Environment
Department of Environmental Engineering
Miljoevej, building 113
DK-2800 Kgs. Lyngby
Denmark
Phone: +45 4525 1600
Fax: +45 4593 2850
e-mail: [email protected]
www.env.dtu.dk
The Department of Environmental Engineering (DTU Environment) conducts
science-based engineering research within four themes:
Water Resource Engineering, Urban Water Engineering,
Residual Resource Engineering and Environmental Chemistry & Microbiology.
Each theme hosts two to four research groups.
The department dates back to 1865, when Ludvig August Colding, the
founder of the , gave the first lecture on sanitary engineering as
response to the cholera epidemics in Copenhagen in the late 1800s.
department
ISBN 978-87-92654-72-4